|
223
|
1 /* vi:set ts=8 sts=4 sw=4:
|
|
|
2 *
|
|
|
3 * VIM - Vi IMproved by Bram Moolenaar
|
|
|
4 *
|
|
|
5 * Do ":help uganda" in Vim to read copying and usage conditions.
|
|
|
6 * Do ":help credits" in Vim to see a list of people who contributed.
|
|
|
7 * See README.txt for an overview of the Vim source code.
|
|
|
8 */
|
|
|
9
|
|
|
10 /*
|
|
|
11 * spell.c: code for spell checking
|
|
226
|
12 *
|
|
300
|
13 * The spell checking mechanism uses a tree (aka trie). Each node in the tree
|
|
|
14 * has a list of bytes that can appear (siblings). For each byte there is a
|
|
|
15 * pointer to the node with the byte that follows in the word (child).
|
|
324
|
16 *
|
|
|
17 * A NUL byte is used where the word may end. The bytes are sorted, so that
|
|
|
18 * binary searching can be used and the NUL bytes are at the start. The
|
|
|
19 * number of possible bytes is stored before the list of bytes.
|
|
|
20 *
|
|
|
21 * The tree uses two arrays: "byts" stores the characters, "idxs" stores
|
|
|
22 * either the next index or flags. The tree starts at index 0. For example,
|
|
|
23 * to lookup "vi" this sequence is followed:
|
|
|
24 * i = 0
|
|
|
25 * len = byts[i]
|
|
|
26 * n = where "v" appears in byts[i + 1] to byts[i + len]
|
|
|
27 * i = idxs[n]
|
|
|
28 * len = byts[i]
|
|
|
29 * n = where "i" appears in byts[i + 1] to byts[i + len]
|
|
|
30 * i = idxs[n]
|
|
|
31 * len = byts[i]
|
|
|
32 * find that byts[i + 1] is 0, idxs[i + 1] has flags for "vi".
|
|
300
|
33 *
|
|
339
|
34 * There are two word trees: one with case-folded words and one with words in
|
|
300
|
35 * original case. The second one is only used for keep-case words and is
|
|
|
36 * usually small.
|
|
|
37 *
|
|
481
|
38 * There is one additional tree for when not all prefixes are applied when
|
|
339
|
39 * generating the .spl file. This tree stores all the possible prefixes, as
|
|
|
40 * if they were words. At each word (prefix) end the prefix nr is stored, the
|
|
|
41 * following word must support this prefix nr. And the condition nr is
|
|
|
42 * stored, used to lookup the condition that the word must match with.
|
|
|
43 *
|
|
300
|
44 * Thanks to Olaf Seibert for providing an example implementation of this tree
|
|
|
45 * and the compression mechanism.
|
|
243
|
46 *
|
|
|
47 * Matching involves checking the caps type: Onecap ALLCAP KeepCap.
|
|
|
48 *
|
|
236
|
49 * Why doesn't Vim use aspell/ispell/myspell/etc.?
|
|
|
50 * See ":help develop-spell".
|
|
|
51 */
|
|
|
52
|
|
469
|
53 /* Use SPELL_PRINTTREE for debugging: dump the word tree after adding a word.
|
|
482
|
54 * Only use it for small word lists! */
|
|
469
|
55 #if 0
|
|
|
56 # define SPELL_PRINTTREE
|
|
482
|
57 #endif
|
|
|
58
|
|
300
|
59 /*
|
|
324
|
60 * Use this to adjust the score after finding suggestions, based on the
|
|
|
61 * suggested word sounding like the bad word. This is much faster than doing
|
|
|
62 * it for every possible suggestion.
|
|
|
63 * Disadvantage: When "the" is typed as "hte" it sounds different and goes
|
|
|
64 * down in the list.
|
|
344
|
65 * Used when 'spellsuggest' is set to "best".
|
|
|
66 */
|
|
|
67 #define RESCORE(word_score, sound_score) ((3 * word_score + sound_score) / 4)
|
|
|
68
|
|
|
69 /*
|
|
339
|
70 * Vim spell file format: <HEADER>
|
|
483
|
71 * <SECTIONS>
|
|
339
|
72 * <LWORDTREE>
|
|
|
73 * <KWORDTREE>
|
|
|
74 * <PREFIXTREE>
|
|
300
|
75 *
|
|
483
|
76 * <HEADER>: <fileID> <versionnr>
|
|
|
77 *
|
|
|
78 * <fileID> 8 bytes "VIMspell"
|
|
|
79 * <versionnr> 1 byte VIMSPELLVERSION
|
|
|
80 *
|
|
|
81 *
|
|
|
82 * Sections make it possible to add information to the .spl file without
|
|
|
83 * making it incompatible with previous versions. There are two kinds of
|
|
|
84 * sections:
|
|
|
85 * 1. Not essential for correct spell checking. E.g. for making suggestions.
|
|
|
86 * These are skipped when not supported.
|
|
|
87 * 2. Optional information, but essential for spell checking when present.
|
|
|
88 * E.g. conditions for affixes. When this section is present but not
|
|
|
89 * supported an error message is given.
|
|
300
|
90 *
|
|
483
|
91 * <SECTIONS>: <section> ... <sectionend>
|
|
|
92 *
|
|
|
93 * <section>: <sectionID> <sectionflags> <sectionlen> (section contents)
|
|
|
94 *
|
|
|
95 * <sectionID> 1 byte number from 0 to 254 identifying the section
|
|
|
96 *
|
|
|
97 * <sectionflags> 1 byte SNF_REQUIRED: this section is required for correct
|
|
|
98 * spell checking
|
|
|
99 *
|
|
|
100 * <sectionlen> 4 bytes length of section contents, MSB first
|
|
|
101 *
|
|
|
102 * <sectionend> 1 byte SN_END
|
|
|
103 *
|
|
|
104 *
|
|
|
105 * sectionID == SN_REGION: <regionname> ...
|
|
|
106 * <regionname> 2 bytes Up to 8 region names: ca, au, etc. Lower case.
|
|
300
|
107 * First <regionname> is region 1.
|
|
|
108 *
|
|
483
|
109 * sectionID == SN_CHARFLAGS: <charflagslen> <charflags>
|
|
|
110 * <folcharslen> <folchars>
|
|
300
|
111 * <charflagslen> 1 byte Number of bytes in <charflags> (should be 128).
|
|
|
112 * <charflags> N bytes List of flags (first one is for character 128):
|
|
324
|
113 * 0x01 word character CF_WORD
|
|
|
114 * 0x02 upper-case character CF_UPPER
|
|
483
|
115 * <folcharslen> 2 bytes Number of bytes in <folchars>.
|
|
|
116 * <folchars> N bytes Folded characters, first one is for character 128.
|
|
300
|
117 *
|
|
483
|
118 * sectionID == SN_MIDWORD: <midword>
|
|
|
119 * <midword> N bytes Characters that are word characters only when used
|
|
366
|
120 * in the middle of a word.
|
|
|
121 *
|
|
483
|
122 * sectionID == SN_PREFCOND: <prefcondcnt> <prefcond> ...
|
|
339
|
123 * <prefcondcnt> 2 bytes Number of <prefcond> items following.
|
|
|
124 * <prefcond> : <condlen> <condstr>
|
|
|
125 * <condlen> 1 byte Length of <condstr>.
|
|
|
126 * <condstr> N bytes Condition for the prefix.
|
|
|
127 *
|
|
483
|
128 * sectionID == SN_REP: <repcount> <rep> ...
|
|
|
129 * <repcount> 2 bytes number of <rep> items, MSB first.
|
|
323
|
130 * <rep> : <repfromlen> <repfrom> <reptolen> <repto>
|
|
483
|
131 * <repfromlen> 1 byte length of <repfrom>
|
|
|
132 * <repfrom> N bytes "from" part of replacement
|
|
|
133 * <reptolen> 1 byte length of <repto>
|
|
|
134 * <repto> N bytes "to" part of replacement
|
|
323
|
135 *
|
|
483
|
136 * sectionID == SN_SAL: <salflags> <salcount> <sal> ...
|
|
|
137 * <salflags> 1 byte flags for soundsalike conversion:
|
|
323
|
138 * SAL_F0LLOWUP
|
|
|
139 * SAL_COLLAPSE
|
|
|
140 * SAL_REM_ACCENTS
|
|
483
|
141 * <salcount> 2 bytes number of <sal> items following
|
|
323
|
142 * <sal> : <salfromlen> <salfrom> <saltolen> <salto>
|
|
483
|
143 * <salfromlen> 1 byte length of <salfrom>
|
|
|
144 * <salfrom> N bytes "from" part of soundsalike
|
|
|
145 * <saltolen> 1 byte length of <salto>
|
|
|
146 * <salto> N bytes "to" part of soundsalike
|
|
323
|
147 *
|
|
483
|
148 * sectionID == SN_SOFO: <sofofromlen> <sofofrom> <sofotolen> <sofoto>
|
|
|
149 * <sofofromlen> 2 bytes length of <sofofrom>
|
|
|
150 * <sofofrom> N bytes "from" part of soundfold
|
|
|
151 * <sofotolen> 2 bytes length of <sofoto>
|
|
|
152 * <sofoto> N bytes "to" part of soundfold
|
|
323
|
153 *
|
|
483
|
154 * sectionID == SN_MAP: <mapstr>
|
|
|
155 * <mapstr> N bytes String with sequences of similar characters,
|
|
323
|
156 * separated by slashes.
|
|
300
|
157 *
|
|
483
|
158 * sectionID == SN_COMPOUND: <compmax> <compminlen> <compsylmax> <compflags>
|
|
|
159 * <compmax> 1 byte Maximum nr of words in compound word.
|
|
|
160 * <compminlen> 1 byte Minimal word length for compounding.
|
|
|
161 * <compsylmax> 1 byte Maximum nr of syllables in compound word.
|
|
|
162 * <compflags> N bytes Flags from COMPOUNDFLAGS items, separated by
|
|
|
163 * slashes.
|
|
|
164 *
|
|
492
|
165 * sectionID == SN_NOBREAK: (empty, its presence is enough)
|
|
|
166 *
|
|
483
|
167 * sectionID == SN_SYLLABLE: <syllable>
|
|
|
168 * <syllable> N bytes String from SYLLABLE item.
|
|
300
|
169 *
|
|
|
170 * <LWORDTREE>: <wordtree>
|
|
|
171 *
|
|
339
|
172 * <KWORDTREE>: <wordtree>
|
|
|
173 *
|
|
|
174 * <PREFIXTREE>: <wordtree>
|
|
|
175 *
|
|
|
176 *
|
|
300
|
177 * <wordtree>: <nodecount> <nodedata> ...
|
|
|
178 *
|
|
|
179 * <nodecount> 4 bytes Number of nodes following. MSB first.
|
|
|
180 *
|
|
|
181 * <nodedata>: <siblingcount> <sibling> ...
|
|
|
182 *
|
|
|
183 * <siblingcount> 1 byte Number of siblings in this node. The siblings
|
|
|
184 * follow in sorted order.
|
|
|
185 *
|
|
339
|
186 * <sibling>: <byte> [ <nodeidx> <xbyte>
|
|
481
|
187 * | <flags> [<flags2>] [<region>] [<affixID>]
|
|
|
188 * | [<pflags>] <affixID> <prefcondnr> ]
|
|
300
|
189 *
|
|
|
190 * <byte> 1 byte Byte value of the sibling. Special cases:
|
|
|
191 * BY_NOFLAGS: End of word without flags and for all
|
|
|
192 * regions.
|
|
481
|
193 * For PREFIXTREE <affixID> and
|
|
366
|
194 * <prefcondnr> follow.
|
|
390
|
195 * BY_FLAGS: End of word, <flags> follow.
|
|
481
|
196 * For PREFIXTREE <pflags>, <affixID>
|
|
455
|
197 * and <prefcondnr> follow.
|
|
390
|
198 * BY_FLAGS2: End of word, <flags> and <flags2>
|
|
|
199 * follow. Not used in PREFIXTREE.
|
|
|
200 * BY_INDEX: Child of sibling is shared, <nodeidx>
|
|
300
|
201 * and <xbyte> follow.
|
|
|
202 *
|
|
|
203 * <nodeidx> 3 bytes Index of child for this sibling, MSB first.
|
|
|
204 *
|
|
|
205 * <xbyte> 1 byte byte value of the sibling.
|
|
|
206 *
|
|
|
207 * <flags> 1 byte bitmask of:
|
|
|
208 * WF_ALLCAP word must have only capitals
|
|
|
209 * WF_ONECAP first char of word must be capital
|
|
389
|
210 * WF_KEEPCAP keep-case word
|
|
|
211 * WF_FIXCAP keep-case word, all caps not allowed
|
|
300
|
212 * WF_RARE rare word
|
|
389
|
213 * WF_BANNED bad word
|
|
300
|
214 * WF_REGION <region> follows
|
|
481
|
215 * WF_AFX <affixID> follows
|
|
300
|
216 *
|
|
390
|
217 * <flags2> 1 byte Only used when there are postponed prefixes.
|
|
|
218 * Bitmask of:
|
|
|
219 * WF_HAS_AFF >> 8 word includes affix
|
|
|
220 *
|
|
455
|
221 * <pflags> 1 byte bitmask of:
|
|
|
222 * WFP_RARE rare prefix
|
|
|
223 * WFP_NC non-combining prefix
|
|
|
224 * WFP_UP letter after prefix made upper case
|
|
|
225 *
|
|
300
|
226 * <region> 1 byte Bitmask for regions in which word is valid. When
|
|
|
227 * omitted it's valid in all regions.
|
|
|
228 * Lowest bit is for region 1.
|
|
|
229 *
|
|
481
|
230 * <affixID> 1 byte ID of affix that can be used with this word. In
|
|
339
|
231 * PREFIXTREE used for the required prefix ID.
|
|
|
232 *
|
|
|
233 * <prefcondnr> 2 bytes Prefix condition number, index in <prefcond> list
|
|
|
234 * from HEADER.
|
|
300
|
235 *
|
|
|
236 * All text characters are in 'encoding', but stored as single bytes.
|
|
|
237 */
|
|
|
238
|
|
223
|
239 #if defined(MSDOS) || defined(WIN16) || defined(WIN32) || defined(_WIN64)
|
|
|
240 # include <io.h> /* for lseek(), must be before vim.h */
|
|
|
241 #endif
|
|
|
242
|
|
|
243 #include "vim.h"
|
|
|
244
|
|
|
245 #if defined(FEAT_SYN_HL) || defined(PROTO)
|
|
|
246
|
|
|
247 #ifdef HAVE_FCNTL_H
|
|
|
248 # include <fcntl.h>
|
|
|
249 #endif
|
|
|
250
|
|
323
|
251 #define MAXWLEN 250 /* Assume max. word len is this many bytes.
|
|
|
252 Some places assume a word length fits in a
|
|
|
253 byte, thus it can't be above 255. */
|
|
226
|
254
|
|
490
|
255 /* Type used for indexes in the word tree need to be at least 4 bytes. If int
|
|
324
|
256 * is 8 bytes we could use something smaller, but what? */
|
|
490
|
257 #if SIZEOF_INT > 3
|
|
324
|
258 typedef int idx_T;
|
|
|
259 #else
|
|
|
260 typedef long idx_T;
|
|
|
261 #endif
|
|
|
262
|
|
|
263 /* Flags used for a word. Only the lowest byte can be used, the region byte
|
|
|
264 * comes above it. */
|
|
300
|
265 #define WF_REGION 0x01 /* region byte follows */
|
|
|
266 #define WF_ONECAP 0x02 /* word with one capital (or all capitals) */
|
|
|
267 #define WF_ALLCAP 0x04 /* word must be all capitals */
|
|
|
268 #define WF_RARE 0x08 /* rare word */
|
|
307
|
269 #define WF_BANNED 0x10 /* bad word */
|
|
481
|
270 #define WF_AFX 0x20 /* affix ID follows */
|
|
389
|
271 #define WF_FIXCAP 0x40 /* keep-case word, allcap not allowed */
|
|
323
|
272 #define WF_KEEPCAP 0x80 /* keep-case word */
|
|
|
273
|
|
390
|
274 /* for <flags2>, shifted up one byte to be used in wn_flags */
|
|
|
275 #define WF_HAS_AFF 0x0100 /* word includes affix */
|
|
|
276
|
|
389
|
277 #define WF_CAPMASK (WF_ONECAP | WF_ALLCAP | WF_KEEPCAP | WF_FIXCAP)
|
|
300
|
278
|
|
455
|
279 /* flags for <pflags> */
|
|
|
280 #define WFP_RARE 0x01 /* rare prefix */
|
|
|
281 #define WFP_NC 0x02 /* prefix is not combining */
|
|
|
282 #define WFP_UP 0x04 /* to-upper prefix */
|
|
|
283
|
|
481
|
284 /* Flags for postponed prefixes. Must be above affixID (one byte)
|
|
390
|
285 * and prefcondnr (two bytes). */
|
|
455
|
286 #define WF_RAREPFX (WFP_RARE << 24) /* in sl_pidxs: flag for rare
|
|
|
287 * postponed prefix */
|
|
|
288 #define WF_PFX_NC (WFP_NC << 24) /* in sl_pidxs: flag for non-combining
|
|
|
289 * postponed prefix */
|
|
|
290 #define WF_PFX_UP (WFP_UP << 24) /* in sl_pidxs: flag for to-upper
|
|
|
291 * postponed prefix */
|
|
390
|
292
|
|
|
293 /* Special byte values for <byte>. Some are only used in the tree for
|
|
|
294 * postponed prefixes, some only in the other trees. This is a bit messy... */
|
|
|
295 #define BY_NOFLAGS 0 /* end of word without flags or region; for
|
|
455
|
296 * postponed prefix: no <pflags> */
|
|
|
297 #define BY_INDEX 1 /* child is shared, index follows */
|
|
|
298 #define BY_FLAGS 2 /* end of word, <flags> byte follows; for
|
|
|
299 * postponed prefix: <pflags> follows */
|
|
|
300 #define BY_FLAGS2 3 /* end of word, <flags> and <flags2> bytes
|
|
|
301 * follow; never used in prefix tree */
|
|
|
302 #define BY_SPECIAL BY_FLAGS2 /* highest special byte value */
|
|
236
|
303
|
|
323
|
304 /* Info from "REP" and "SAL" entries in ".aff" file used in si_rep, sl_rep,
|
|
344
|
305 * and si_sal. Not for sl_sal!
|
|
323
|
306 * One replacement: from "ft_from" to "ft_to". */
|
|
|
307 typedef struct fromto_S
|
|
236
|
308 {
|
|
323
|
309 char_u *ft_from;
|
|
|
310 char_u *ft_to;
|
|
|
311 } fromto_T;
|
|
236
|
312
|
|
344
|
313 /* Info from "SAL" entries in ".aff" file used in sl_sal.
|
|
|
314 * The info is split for quick processing by spell_soundfold().
|
|
|
315 * Note that "sm_oneof" and "sm_rules" point into sm_lead. */
|
|
|
316 typedef struct salitem_S
|
|
|
317 {
|
|
|
318 char_u *sm_lead; /* leading letters */
|
|
|
319 int sm_leadlen; /* length of "sm_lead" */
|
|
375
|
320 char_u *sm_oneof; /* letters from () or NULL */
|
|
344
|
321 char_u *sm_rules; /* rules like ^, $, priority */
|
|
|
322 char_u *sm_to; /* replacement. */
|
|
372
|
323 #ifdef FEAT_MBYTE
|
|
|
324 int *sm_lead_w; /* wide character copy of "sm_lead" */
|
|
375
|
325 int *sm_oneof_w; /* wide character copy of "sm_oneof" */
|
|
372
|
326 int *sm_to_w; /* wide character copy of "sm_to" */
|
|
|
327 #endif
|
|
344
|
328 } salitem_T;
|
|
|
329
|
|
375
|
330 #ifdef FEAT_MBYTE
|
|
|
331 typedef int salfirst_T;
|
|
|
332 #else
|
|
|
333 typedef short salfirst_T;
|
|
|
334 #endif
|
|
|
335
|
|
483
|
336 /* Values for SP_*ERROR are negative, positive values are used by
|
|
|
337 * read_cnt_string(). */
|
|
|
338 #define SP_TRUNCERROR -1 /* spell file truncated error */
|
|
|
339 #define SP_FORMERROR -2 /* format error in spell file */
|
|
|
340 #define SP_ERROR -3 /* other error while reading spell file */
|
|
|
341
|
|
236
|
342 /*
|
|
243
|
343 * Structure used to store words and other info for one language, loaded from
|
|
|
344 * a .spl file.
|
|
300
|
345 * The main access is through the tree in "sl_fbyts/sl_fidxs", storing the
|
|
|
346 * case-folded words. "sl_kbyts/sl_kidxs" is for keep-case words.
|
|
|
347 *
|
|
|
348 * The "byts" array stores the possible bytes in each tree node, preceded by
|
|
|
349 * the number of possible bytes, sorted on byte value:
|
|
|
350 * <len> <byte1> <byte2> ...
|
|
|
351 * The "idxs" array stores the index of the child node corresponding to the
|
|
|
352 * byte in "byts".
|
|
|
353 * Exception: when the byte is zero, the word may end here and "idxs" holds
|
|
481
|
354 * the flags, region mask and affixID for the word. There may be several
|
|
|
355 * zeros in sequence for alternative flag/region/affixID combinations.
|
|
236
|
356 */
|
|
|
357 typedef struct slang_S slang_T;
|
|
|
358 struct slang_S
|
|
|
359 {
|
|
|
360 slang_T *sl_next; /* next language */
|
|
|
361 char_u *sl_name; /* language name "en", "en.rare", "nl", etc. */
|
|
310
|
362 char_u *sl_fname; /* name of .spl file */
|
|
323
|
363 int sl_add; /* TRUE if it's a .add file. */
|
|
339
|
364
|
|
300
|
365 char_u *sl_fbyts; /* case-folded word bytes */
|
|
324
|
366 idx_T *sl_fidxs; /* case-folded word indexes */
|
|
300
|
367 char_u *sl_kbyts; /* keep-case word bytes */
|
|
324
|
368 idx_T *sl_kidxs; /* keep-case word indexes */
|
|
339
|
369 char_u *sl_pbyts; /* prefix tree word bytes */
|
|
|
370 idx_T *sl_pidxs; /* prefix tree word indexes */
|
|
|
371
|
|
236
|
372 char_u sl_regions[17]; /* table with up to 8 region names plus NUL */
|
|
323
|
373
|
|
376
|
374 char_u *sl_midword; /* MIDWORD string or NULL */
|
|
|
375
|
|
483
|
376 int sl_compmax; /* COMPOUNDMAX (default: MAXWLEN) */
|
|
|
377 int sl_compminlen; /* COMPOUNDMIN (default: MAXWLEN) */
|
|
|
378 int sl_compsylmax; /* COMPOUNDSYLMAX (default: MAXWLEN) */
|
|
|
379 regprog_T *sl_compprog; /* COMPOUNDFLAGS turned into a regexp progrm
|
|
|
380 * (NULL when no compounding) */
|
|
|
381 char_u *sl_compstartflags; /* flags for first compound word */
|
|
485
|
382 char_u *sl_compallflags; /* all flags for compound words */
|
|
492
|
383 char_u sl_nobreak; /* When TRUE: no spaces between words */
|
|
483
|
384 char_u *sl_syllable; /* SYLLABLE repeatable chars or NULL */
|
|
|
385 garray_T sl_syl_items; /* syllable items */
|
|
481
|
386
|
|
339
|
387 int sl_prefixcnt; /* number of items in "sl_prefprog" */
|
|
|
388 regprog_T **sl_prefprog; /* table with regprogs for prefixes */
|
|
|
389
|
|
323
|
390 garray_T sl_rep; /* list of fromto_T entries from REP lines */
|
|
|
391 short sl_rep_first[256]; /* indexes where byte first appears, -1 if
|
|
|
392 there is none */
|
|
344
|
393 garray_T sl_sal; /* list of salitem_T entries from SAL lines */
|
|
375
|
394 salfirst_T sl_sal_first[256]; /* indexes where byte first appears, -1 if
|
|
323
|
395 there is none */
|
|
375
|
396 int sl_sofo; /* SOFOFROM and SOFOTO instead of SAL items:
|
|
|
397 * "sl_sal_first" maps chars, when has_mbyte
|
|
|
398 * "sl_sal" is a list of wide char lists. */
|
|
323
|
399 int sl_followup; /* SAL followup */
|
|
|
400 int sl_collapse; /* SAL collapse_result */
|
|
|
401 int sl_rem_accents; /* SAL remove_accents */
|
|
330
|
402 int sl_has_map; /* TRUE if there is a MAP line */
|
|
|
403 #ifdef FEAT_MBYTE
|
|
|
404 hashtab_T sl_map_hash; /* MAP for multi-byte chars */
|
|
|
405 int sl_map_array[256]; /* MAP for first 256 chars */
|
|
|
406 #else
|
|
|
407 char_u sl_map_array[256]; /* MAP for first 256 chars */
|
|
|
408 #endif
|
|
236
|
409 };
|
|
|
410
|
|
243
|
411 /* First language that is loaded, start of the linked list of loaded
|
|
|
412 * languages. */
|
|
236
|
413 static slang_T *first_lang = NULL;
|
|
|
414
|
|
323
|
415 /* Flags used in .spl file for soundsalike flags. */
|
|
|
416 #define SAL_F0LLOWUP 1
|
|
|
417 #define SAL_COLLAPSE 2
|
|
|
418 #define SAL_REM_ACCENTS 4
|
|
|
419
|
|
236
|
420 /*
|
|
|
421 * Structure used in "b_langp", filled from 'spelllang'.
|
|
|
422 */
|
|
|
423 typedef struct langp_S
|
|
|
424 {
|
|
|
425 slang_T *lp_slang; /* info for this language (NULL for last one) */
|
|
|
426 int lp_region; /* bitmask for region or REGION_ALL */
|
|
|
427 } langp_T;
|
|
|
428
|
|
|
429 #define LANGP_ENTRY(ga, i) (((langp_T *)(ga).ga_data) + (i))
|
|
|
430
|
|
307
|
431 #define REGION_ALL 0xff /* word valid in all regions */
|
|
|
432
|
|
483
|
433 #define VIMSPELLMAGIC "VIMspell" /* string at start of Vim spell file */
|
|
|
434 #define VIMSPELLMAGICL 8
|
|
|
435 #define VIMSPELLVERSION 50
|
|
|
436
|
|
|
437 /* Section IDs. Only renumber them when VIMSPELLVERSION changes! */
|
|
|
438 #define SN_REGION 0 /* <regionname> section */
|
|
|
439 #define SN_CHARFLAGS 1 /* charflags section */
|
|
|
440 #define SN_MIDWORD 2 /* <midword> section */
|
|
|
441 #define SN_PREFCOND 3 /* <prefcond> section */
|
|
|
442 #define SN_REP 4 /* REP items section */
|
|
|
443 #define SN_SAL 5 /* SAL items section */
|
|
|
444 #define SN_SOFO 6 /* soundfolding section */
|
|
|
445 #define SN_MAP 7 /* MAP items section */
|
|
|
446 #define SN_COMPOUND 8 /* compound words section */
|
|
|
447 #define SN_SYLLABLE 9 /* syllable section */
|
|
492
|
448 #define SN_NOBREAK 10 /* NOBREAK section */
|
|
483
|
449 #define SN_END 255 /* end of sections */
|
|
|
450
|
|
|
451 #define SNF_REQUIRED 1 /* <sectionflags>: required section */
|
|
|
452
|
|
307
|
453 /* Result values. Lower number is accepted over higher one. */
|
|
|
454 #define SP_BANNED -1
|
|
236
|
455 #define SP_OK 0
|
|
307
|
456 #define SP_RARE 1
|
|
|
457 #define SP_LOCAL 2
|
|
|
458 #define SP_BAD 3
|
|
236
|
459
|
|
381
|
460 /* file used for "zG" and "zW" */
|
|
385
|
461 static char_u *int_wordlist = NULL;
|
|
381
|
462
|
|
236
|
463 /*
|
|
323
|
464 * Information used when looking for suggestions.
|
|
|
465 */
|
|
|
466 typedef struct suginfo_S
|
|
|
467 {
|
|
|
468 garray_T su_ga; /* suggestions, contains "suggest_T" */
|
|
344
|
469 int su_maxcount; /* max. number of suggestions displayed */
|
|
323
|
470 int su_maxscore; /* maximum score for adding to su_ga */
|
|
344
|
471 garray_T su_sga; /* like su_ga, sound-folded scoring */
|
|
323
|
472 char_u *su_badptr; /* start of bad word in line */
|
|
|
473 int su_badlen; /* length of detected bad word in line */
|
|
346
|
474 int su_badflags; /* caps flags for bad word */
|
|
323
|
475 char_u su_badword[MAXWLEN]; /* bad word truncated at su_badlen */
|
|
|
476 char_u su_fbadword[MAXWLEN]; /* su_badword case-folded */
|
|
|
477 hashtab_T su_banned; /* table with banned words */
|
|
|
478 } suginfo_T;
|
|
|
479
|
|
|
480 /* One word suggestion. Used in "si_ga". */
|
|
|
481 typedef struct suggest_S
|
|
|
482 {
|
|
|
483 char_u *st_word; /* suggested word, allocated string */
|
|
|
484 int st_orglen; /* length of replaced text */
|
|
|
485 int st_score; /* lower is better */
|
|
344
|
486 int st_altscore; /* used when st_score compares equal */
|
|
|
487 int st_salscore; /* st_score is for soundalike */
|
|
324
|
488 int st_had_bonus; /* bonus already included in score */
|
|
323
|
489 } suggest_T;
|
|
|
490
|
|
344
|
491 #define SUG(ga, i) (((suggest_T *)(ga).ga_data)[i])
|
|
323
|
492
|
|
324
|
493 /* Number of suggestions kept when cleaning up. When rescore_suggestions() is
|
|
|
494 * called the score may change, thus we need to keep more than what is
|
|
|
495 * displayed. */
|
|
351
|
496 #define SUG_CLEAN_COUNT(su) ((su)->su_maxcount < 50 ? 50 : (su)->su_maxcount)
|
|
324
|
497
|
|
|
498 /* Threshold for sorting and cleaning up suggestions. Don't want to keep lots
|
|
|
499 * of suggestions that are not going to be displayed. */
|
|
344
|
500 #define SUG_MAX_COUNT(su) ((su)->su_maxcount + 50)
|
|
323
|
501
|
|
|
502 /* score for various changes */
|
|
344
|
503 #define SCORE_SPLIT 149 /* split bad word */
|
|
323
|
504 #define SCORE_ICASE 52 /* slightly different case */
|
|
483
|
505 #define SCORE_REGION 200 /* word is for different region */
|
|
323
|
506 #define SCORE_RARE 180 /* rare word */
|
|
|
507 #define SCORE_SWAP 90 /* swap two characters */
|
|
|
508 #define SCORE_SWAP3 110 /* swap two characters in three */
|
|
|
509 #define SCORE_REP 87 /* REP replacement */
|
|
|
510 #define SCORE_SUBST 93 /* substitute a character */
|
|
|
511 #define SCORE_SIMILAR 33 /* substitute a similar character */
|
|
479
|
512 #define SCORE_SUBCOMP 33 /* substitute a composing character */
|
|
324
|
513 #define SCORE_DEL 94 /* delete a character */
|
|
358
|
514 #define SCORE_DELDUP 64 /* delete a duplicated character */
|
|
479
|
515 #define SCORE_DELCOMP 28 /* delete a composing character */
|
|
324
|
516 #define SCORE_INS 96 /* insert a character */
|
|
358
|
517 #define SCORE_INSDUP 66 /* insert a duplicate character */
|
|
475
|
518 #define SCORE_INSCOMP 30 /* insert a composing character */
|
|
366
|
519 #define SCORE_NONWORD 103 /* change non-word to word char */
|
|
323
|
520
|
|
372
|
521 #define SCORE_FILE 30 /* suggestion from a file */
|
|
323
|
522 #define SCORE_MAXINIT 350 /* Initial maximum score: higher == slower.
|
|
|
523 * 350 allows for about three changes. */
|
|
344
|
524
|
|
|
525 #define SCORE_BIG SCORE_INS * 3 /* big difference */
|
|
323
|
526 #define SCORE_MAXMAX 999999 /* accept any score */
|
|
|
527
|
|
|
528 /*
|
|
236
|
529 * Structure to store info for word matching.
|
|
|
530 */
|
|
|
531 typedef struct matchinf_S
|
|
|
532 {
|
|
|
533 langp_T *mi_lp; /* info for language and region */
|
|
243
|
534
|
|
|
535 /* pointers to original text to be checked */
|
|
236
|
536 char_u *mi_word; /* start of word being checked */
|
|
339
|
537 char_u *mi_end; /* end of matching word so far */
|
|
243
|
538 char_u *mi_fend; /* next char to be added to mi_fword */
|
|
300
|
539 char_u *mi_cend; /* char after what was used for
|
|
|
540 mi_capflags */
|
|
243
|
541
|
|
|
542 /* case-folded text */
|
|
|
543 char_u mi_fword[MAXWLEN + 1]; /* mi_word case-folded */
|
|
300
|
544 int mi_fwordlen; /* nr of valid bytes in mi_fword */
|
|
243
|
545
|
|
339
|
546 /* for when checking word after a prefix */
|
|
|
547 int mi_prefarridx; /* index in sl_pidxs with list of
|
|
481
|
548 affixID/condition */
|
|
339
|
549 int mi_prefcnt; /* number of entries at mi_prefarridx */
|
|
|
550 int mi_prefixlen; /* byte length of prefix */
|
|
455
|
551 #ifdef FEAT_MBYTE
|
|
|
552 int mi_cprefixlen; /* byte length of prefix in original
|
|
|
553 case */
|
|
|
554 #else
|
|
|
555 # define mi_cprefixlen mi_prefixlen /* it's the same value */
|
|
|
556 #endif
|
|
339
|
557
|
|
481
|
558 /* for when checking a compound word */
|
|
|
559 int mi_compoff; /* start of following word offset */
|
|
483
|
560 char_u mi_compflags[MAXWLEN]; /* flags for compound words used */
|
|
|
561 int mi_complen; /* nr of compound words used */
|
|
481
|
562
|
|
243
|
563 /* others */
|
|
236
|
564 int mi_result; /* result so far: SP_BAD, SP_OK, etc. */
|
|
300
|
565 int mi_capflags; /* WF_ONECAP WF_ALLCAP WF_KEEPCAP */
|
|
376
|
566 buf_T *mi_buf; /* buffer being checked */
|
|
492
|
567
|
|
|
568 /* for NOBREAK */
|
|
|
569 int mi_result2; /* "mi_resul" without following word */
|
|
|
570 char_u *mi_end2; /* "mi_end" without following word */
|
|
236
|
571 } matchinf_T;
|
|
|
572
|
|
307
|
573 /*
|
|
|
574 * The tables used for recognizing word characters according to spelling.
|
|
|
575 * These are only used for the first 256 characters of 'encoding'.
|
|
|
576 */
|
|
|
577 typedef struct spelltab_S
|
|
|
578 {
|
|
|
579 char_u st_isw[256]; /* flags: is word char */
|
|
|
580 char_u st_isu[256]; /* flags: is uppercase char */
|
|
|
581 char_u st_fold[256]; /* chars: folded case */
|
|
324
|
582 char_u st_upper[256]; /* chars: upper case */
|
|
307
|
583 } spelltab_T;
|
|
|
584
|
|
|
585 static spelltab_T spelltab;
|
|
|
586 static int did_set_spelltab;
|
|
|
587
|
|
324
|
588 #define CF_WORD 0x01
|
|
|
589 #define CF_UPPER 0x02
|
|
307
|
590
|
|
|
591 static void clear_spell_chartab __ARGS((spelltab_T *sp));
|
|
|
592 static int set_spell_finish __ARGS((spelltab_T *new_st));
|
|
376
|
593 static int spell_iswordp __ARGS((char_u *p, buf_T *buf));
|
|
|
594 static int spell_iswordp_nmw __ARGS((char_u *p));
|
|
|
595 #ifdef FEAT_MBYTE
|
|
|
596 static int spell_iswordp_w __ARGS((int *p, buf_T *buf));
|
|
|
597 #endif
|
|
483
|
598 static int write_spell_prefcond __ARGS((FILE *fd, garray_T *gap));
|
|
307
|
599
|
|
|
600 /*
|
|
344
|
601 * For finding suggestions: At each node in the tree these states are tried:
|
|
330
|
602 */
|
|
|
603 typedef enum
|
|
|
604 {
|
|
344
|
605 STATE_START = 0, /* At start of node check for NUL bytes (goodword
|
|
|
606 * ends); if badword ends there is a match, otherwise
|
|
|
607 * try splitting word. */
|
|
375
|
608 STATE_NOPREFIX, /* try without prefix */
|
|
344
|
609 STATE_SPLITUNDO, /* Undo splitting. */
|
|
330
|
610 STATE_ENDNUL, /* Past NUL bytes at start of the node. */
|
|
|
611 STATE_PLAIN, /* Use each byte of the node. */
|
|
|
612 STATE_DEL, /* Delete a byte from the bad word. */
|
|
|
613 STATE_INS, /* Insert a byte in the bad word. */
|
|
|
614 STATE_SWAP, /* Swap two bytes. */
|
|
344
|
615 STATE_UNSWAP, /* Undo swap two characters. */
|
|
|
616 STATE_SWAP3, /* Swap two characters over three. */
|
|
|
617 STATE_UNSWAP3, /* Undo Swap two characters over three. */
|
|
|
618 STATE_UNROT3L, /* Undo rotate three characters left */
|
|
|
619 STATE_UNROT3R, /* Undo rotate three characters right */
|
|
330
|
620 STATE_REP_INI, /* Prepare for using REP items. */
|
|
|
621 STATE_REP, /* Use matching REP items from the .aff file. */
|
|
|
622 STATE_REP_UNDO, /* Undo a REP item replacement. */
|
|
|
623 STATE_FINAL /* End of this node. */
|
|
|
624 } state_T;
|
|
|
625
|
|
|
626 /*
|
|
346
|
627 * Struct to keep the state at each level in suggest_try_change().
|
|
323
|
628 */
|
|
|
629 typedef struct trystate_S
|
|
|
630 {
|
|
330
|
631 state_T ts_state; /* state at this level, STATE_ */
|
|
323
|
632 int ts_score; /* score */
|
|
344
|
633 idx_T ts_arridx; /* index in tree array, start of node */
|
|
330
|
634 short ts_curi; /* index in list of child nodes */
|
|
|
635 char_u ts_fidx; /* index in fword[], case-folded bad word */
|
|
|
636 char_u ts_fidxtry; /* ts_fidx at which bytes may be changed */
|
|
|
637 char_u ts_twordlen; /* valid length of tword[] */
|
|
482
|
638 char_u ts_prefixdepth; /* stack depth for end of prefix or
|
|
485
|
639 * PFD_PREFIXTREE or PFD_NOPREFIX */
|
|
|
640 char_u ts_flags; /* TSF_ flags */
|
|
330
|
641 #ifdef FEAT_MBYTE
|
|
|
642 char_u ts_tcharlen; /* number of bytes in tword character */
|
|
|
643 char_u ts_tcharidx; /* current byte index in tword character */
|
|
|
644 char_u ts_isdiff; /* DIFF_ values */
|
|
|
645 char_u ts_fcharstart; /* index in fword where badword char started */
|
|
|
646 #endif
|
|
483
|
647 char_u ts_prewordlen; /* length of word in "preword[]" */
|
|
|
648 char_u ts_splitoff; /* index in "tword" after last split */
|
|
492
|
649 char_u ts_splitfidx; /* "ts_fidx" at word split */
|
|
483
|
650 char_u ts_complen; /* nr of compound words used */
|
|
485
|
651 char_u ts_compsplit; /* index for "compflags" where word was spit */
|
|
346
|
652 char_u ts_save_badflags; /* su_badflags saved here */
|
|
323
|
653 } trystate_T;
|
|
|
654
|
|
330
|
655 /* values for ts_isdiff */
|
|
|
656 #define DIFF_NONE 0 /* no different byte (yet) */
|
|
|
657 #define DIFF_YES 1 /* different byte found */
|
|
|
658 #define DIFF_INSERT 2 /* inserting character */
|
|
|
659
|
|
485
|
660 /* values for ts_flags */
|
|
|
661 #define TSF_PREFIXOK 1 /* already checked that prefix is OK */
|
|
|
662 #define TSF_DIDSPLIT 2 /* tried split at this point */
|
|
|
663
|
|
375
|
664 /* special values ts_prefixdepth */
|
|
485
|
665 #define PFD_NOPREFIX 0xff /* not using prefixes */
|
|
482
|
666 #define PFD_PREFIXTREE 0xfe /* walking through the prefix tree */
|
|
485
|
667 #define PFD_NOTSPECIAL 0xfd /* first value that's not special */
|
|
375
|
668
|
|
339
|
669 /* mode values for find_word */
|
|
481
|
670 #define FIND_FOLDWORD 0 /* find word case-folded */
|
|
|
671 #define FIND_KEEPWORD 1 /* find keep-case word */
|
|
|
672 #define FIND_PREFIX 2 /* find word after prefix */
|
|
|
673 #define FIND_COMPOUND 3 /* find case-folded compound word */
|
|
|
674 #define FIND_KEEPCOMPOUND 4 /* find keep-case compound word */
|
|
339
|
675
|
|
236
|
676 static slang_T *slang_alloc __ARGS((char_u *lang));
|
|
|
677 static void slang_free __ARGS((slang_T *lp));
|
|
310
|
678 static void slang_clear __ARGS((slang_T *lp));
|
|
339
|
679 static void find_word __ARGS((matchinf_T *mip, int mode));
|
|
483
|
680 static int can_compound __ARGS((slang_T *slang, char_u *word, char_u *flags));
|
|
455
|
681 static int valid_word_prefix __ARGS((int totprefcnt, int arridx, int flags, char_u *word, slang_T *slang, int cond_req));
|
|
485
|
682 static void find_prefix __ARGS((matchinf_T *mip, int mode));
|
|
339
|
683 static int fold_more __ARGS((matchinf_T *mip));
|
|
389
|
684 static int spell_valid_case __ARGS((int wordflags, int treeflags));
|
|
351
|
685 static int no_spell_checking __ARGS((void));
|
|
307
|
686 static void spell_load_lang __ARGS((char_u *lang));
|
|
310
|
687 static char_u *spell_enc __ARGS((void));
|
|
385
|
688 static void int_wordlist_spl __ARGS((char_u *fname));
|
|
310
|
689 static void spell_load_cb __ARGS((char_u *fname, void *cookie));
|
|
323
|
690 static slang_T *spell_load_file __ARGS((char_u *fname, char_u *lang, slang_T *old_lp, int silent));
|
|
389
|
691 static char_u *read_cnt_string __ARGS((FILE *fd, int cnt_bytes, int *lenp));
|
|
483
|
692 static char_u *read_string __ARGS((FILE *fd, int cnt));
|
|
|
693 static int read_region_section __ARGS((FILE *fd, slang_T *slang, int len));
|
|
|
694 static int read_charflags_section __ARGS((FILE *fd));
|
|
|
695 static int read_prefcond_section __ARGS((FILE *fd, slang_T *lp));
|
|
|
696 static int read_rep_section __ARGS((FILE *fd, slang_T *slang));
|
|
|
697 static int read_sal_section __ARGS((FILE *fd, slang_T *slang));
|
|
|
698 static int read_sofo_section __ARGS((FILE *fd, slang_T *slang));
|
|
|
699 static int read_compound __ARGS((FILE *fd, slang_T *slang, int len));
|
|
|
700 static int init_syl_tab __ARGS((slang_T *slang));
|
|
|
701 static int count_syllables __ARGS((slang_T *slang, char_u *word));
|
|
381
|
702 static int set_sofo __ARGS((slang_T *lp, char_u *from, char_u *to));
|
|
|
703 static void set_sal_first __ARGS((slang_T *lp));
|
|
372
|
704 #ifdef FEAT_MBYTE
|
|
|
705 static int *mb_str2wide __ARGS((char_u *s));
|
|
|
706 #endif
|
|
339
|
707 static idx_T read_tree __ARGS((FILE *fd, char_u *byts, idx_T *idxs, int maxidx, int startidx, int prefixtree, int maxprefcondnr));
|
|
376
|
708 static void clear_midword __ARGS((buf_T *buf));
|
|
|
709 static void use_midword __ARGS((slang_T *lp, buf_T *buf));
|
|
236
|
710 static int find_region __ARGS((char_u *rp, char_u *region));
|
|
|
711 static int captype __ARGS((char_u *word, char_u *end));
|
|
474
|
712 static int badword_captype __ARGS((char_u *word, char_u *end));
|
|
323
|
713 static void spell_reload_one __ARGS((char_u *fname, int added_word));
|
|
483
|
714 static void set_spell_charflags __ARGS((char_u *flags, int cnt, char_u *upp));
|
|
307
|
715 static int set_spell_chartab __ARGS((char_u *fol, char_u *low, char_u *upp));
|
|
|
716 static int spell_casefold __ARGS((char_u *p, int len, char_u *buf, int buflen));
|
|
475
|
717 static int check_need_cap __ARGS((linenr_T lnum, colnr_T col));
|
|
388
|
718 static void spell_find_suggest __ARGS((char_u *badptr, suginfo_T *su, int maxcount, int banbadword, int need_cap));
|
|
372
|
719 #ifdef FEAT_EVAL
|
|
|
720 static void spell_suggest_expr __ARGS((suginfo_T *su, char_u *expr));
|
|
|
721 #endif
|
|
|
722 static void spell_suggest_file __ARGS((suginfo_T *su, char_u *fname));
|
|
|
723 static void spell_suggest_intern __ARGS((suginfo_T *su));
|
|
344
|
724 static void spell_find_cleanup __ARGS((suginfo_T *su));
|
|
324
|
725 static void onecap_copy __ARGS((char_u *word, char_u *wcopy, int upper));
|
|
344
|
726 static void allcap_copy __ARGS((char_u *word, char_u *wcopy));
|
|
346
|
727 static void suggest_try_special __ARGS((suginfo_T *su));
|
|
|
728 static void suggest_try_change __ARGS((suginfo_T *su));
|
|
323
|
729 static int try_deeper __ARGS((suginfo_T *su, trystate_T *stack, int depth, int score_add));
|
|
455
|
730 #ifdef FEAT_MBYTE
|
|
|
731 static int nofold_len __ARGS((char_u *fword, int flen, char_u *word));
|
|
|
732 #endif
|
|
323
|
733 static void find_keepcap_word __ARGS((slang_T *slang, char_u *fword, char_u *kword));
|
|
344
|
734 static void score_comp_sal __ARGS((suginfo_T *su));
|
|
|
735 static void score_combine __ARGS((suginfo_T *su));
|
|
351
|
736 static int stp_sal_score __ARGS((suggest_T *stp, suginfo_T *su, slang_T *slang, char_u *badsound));
|
|
346
|
737 static void suggest_try_soundalike __ARGS((suginfo_T *su));
|
|
323
|
738 static void make_case_word __ARGS((char_u *fword, char_u *cword, int flags));
|
|
330
|
739 static void set_map_str __ARGS((slang_T *lp, char_u *map));
|
|
323
|
740 static int similar_chars __ARGS((slang_T *slang, int c1, int c2));
|
|
351
|
741 static void add_suggestion __ARGS((suginfo_T *su, garray_T *gap, char_u *goodword, int badlen, int score, int altscore, int had_bonus));
|
|
323
|
742 static void add_banned __ARGS((suginfo_T *su, char_u *word));
|
|
|
743 static int was_banned __ARGS((suginfo_T *su, char_u *word));
|
|
|
744 static void free_banned __ARGS((suginfo_T *su));
|
|
324
|
745 static void rescore_suggestions __ARGS((suginfo_T *su));
|
|
344
|
746 static int cleanup_suggestions __ARGS((garray_T *gap, int maxscore, int keep));
|
|
375
|
747 static void spell_soundfold __ARGS((slang_T *slang, char_u *inword, int folded, char_u *res));
|
|
|
748 static void spell_soundfold_sofo __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
|
749 static void spell_soundfold_sal __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
372
|
750 #ifdef FEAT_MBYTE
|
|
375
|
751 static void spell_soundfold_wsal __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
372
|
752 #endif
|
|
344
|
753 static int soundalike_score __ARGS((char_u *goodsound, char_u *badsound));
|
|
323
|
754 static int spell_edit_score __ARGS((char_u *badword, char_u *goodword));
|
|
351
|
755 static void dump_word __ARGS((char_u *word, int round, int flags, linenr_T lnum));
|
|
|
756 static linenr_T apply_prefixes __ARGS((slang_T *slang, char_u *word, int round, int flags, linenr_T startlnum));
|
|
323
|
757
|
|
324
|
758 /*
|
|
|
759 * Use our own character-case definitions, because the current locale may
|
|
|
760 * differ from what the .spl file uses.
|
|
|
761 * These must not be called with negative number!
|
|
|
762 */
|
|
|
763 #ifndef FEAT_MBYTE
|
|
|
764 /* Non-multi-byte implementation. */
|
|
|
765 # define SPELL_TOFOLD(c) ((c) < 256 ? spelltab.st_fold[c] : (c))
|
|
|
766 # define SPELL_TOUPPER(c) ((c) < 256 ? spelltab.st_upper[c] : (c))
|
|
|
767 # define SPELL_ISUPPER(c) ((c) < 256 ? spelltab.st_isu[c] : FALSE)
|
|
|
768 #else
|
|
447
|
769 # if defined(HAVE_WCHAR_H)
|
|
|
770 # include <wchar.h> /* for towupper() and towlower() */
|
|
|
771 # endif
|
|
324
|
772 /* Multi-byte implementation. For Unicode we can call utf_*(), but don't do
|
|
|
773 * that for ASCII, because we don't want to use 'casemap' here. Otherwise use
|
|
|
774 * the "w" library function for characters above 255 if available. */
|
|
|
775 # ifdef HAVE_TOWLOWER
|
|
|
776 # define SPELL_TOFOLD(c) (enc_utf8 && (c) >= 128 ? utf_fold(c) \
|
|
|
777 : (c) < 256 ? spelltab.st_fold[c] : towlower(c))
|
|
|
778 # else
|
|
|
779 # define SPELL_TOFOLD(c) (enc_utf8 && (c) >= 128 ? utf_fold(c) \
|
|
|
780 : (c) < 256 ? spelltab.st_fold[c] : (c))
|
|
|
781 # endif
|
|
|
782
|
|
|
783 # ifdef HAVE_TOWUPPER
|
|
|
784 # define SPELL_TOUPPER(c) (enc_utf8 && (c) >= 128 ? utf_toupper(c) \
|
|
|
785 : (c) < 256 ? spelltab.st_upper[c] : towupper(c))
|
|
|
786 # else
|
|
|
787 # define SPELL_TOUPPER(c) (enc_utf8 && (c) >= 128 ? utf_toupper(c) \
|
|
|
788 : (c) < 256 ? spelltab.st_upper[c] : (c))
|
|
|
789 # endif
|
|
|
790
|
|
|
791 # ifdef HAVE_ISWUPPER
|
|
|
792 # define SPELL_ISUPPER(c) (enc_utf8 && (c) >= 128 ? utf_isupper(c) \
|
|
|
793 : (c) < 256 ? spelltab.st_isu[c] : iswupper(c))
|
|
|
794 # else
|
|
|
795 # define SPELL_ISUPPER(c) (enc_utf8 && (c) >= 128 ? utf_isupper(c) \
|
|
474
|
796 : (c) < 256 ? spelltab.st_isu[c] : (FALSE))
|
|
324
|
797 # endif
|
|
|
798 #endif
|
|
|
799
|
|
307
|
800
|
|
|
801 static char *e_format = N_("E759: Format error in spell file");
|
|
381
|
802 static char *e_spell_trunc = N_("E758: Truncated spell file");
|
|
482
|
803 static char *e_afftrailing = N_("Trailing text in %s line %d: %s");
|
|
469
|
804 static char *msg_compressing = N_("Compressing word tree...");
|
|
236
|
805
|
|
|
806 /*
|
|
|
807 * Main spell-checking function.
|
|
300
|
808 * "ptr" points to a character that could be the start of a word.
|
|
236
|
809 * "*attrp" is set to the attributes for a badly spelled word. For a non-word
|
|
|
810 * or when it's OK it remains unchanged.
|
|
|
811 * This must only be called when 'spelllang' is not empty.
|
|
323
|
812 *
|
|
385
|
813 * "capcol" is used to check for a Capitalised word after the end of a
|
|
|
814 * sentence. If it's zero then perform the check. Return the column where to
|
|
|
815 * check next, or -1 when no sentence end was found. If it's NULL then don't
|
|
|
816 * worry.
|
|
323
|
817 *
|
|
236
|
818 * Returns the length of the word in bytes, also when it's OK, so that the
|
|
|
819 * caller can skip over the word.
|
|
|
820 */
|
|
|
821 int
|
|
385
|
822 spell_check(wp, ptr, attrp, capcol)
|
|
236
|
823 win_T *wp; /* current window */
|
|
|
824 char_u *ptr;
|
|
|
825 int *attrp;
|
|
385
|
826 int *capcol; /* column to check for Capital */
|
|
236
|
827 {
|
|
|
828 matchinf_T mi; /* Most things are put in "mi" so that it can
|
|
|
829 be passed to functions quickly. */
|
|
344
|
830 int nrlen = 0; /* found a number first */
|
|
385
|
831 int c;
|
|
483
|
832 int wrongcaplen = 0;
|
|
236
|
833
|
|
307
|
834 /* A word never starts at a space or a control character. Return quickly
|
|
|
835 * then, skipping over the character. */
|
|
|
836 if (*ptr <= ' ')
|
|
|
837 return 1;
|
|
483
|
838 vim_memset(&mi, 0, sizeof(matchinf_T));
|
|
236
|
839
|
|
344
|
840 /* A number is always OK. Also skip hexadecimal numbers 0xFF99 and
|
|
346
|
841 * 0X99FF. But when a word character follows do check spelling to find
|
|
|
842 * "3GPP". */
|
|
300
|
843 if (*ptr >= '0' && *ptr <= '9')
|
|
|
844 {
|
|
316
|
845 if (*ptr == '0' && (ptr[1] == 'x' || ptr[1] == 'X'))
|
|
|
846 mi.mi_end = skiphex(ptr + 2);
|
|
300
|
847 else
|
|
|
848 {
|
|
344
|
849 mi.mi_end = skipdigits(ptr);
|
|
|
850 nrlen = mi.mi_end - ptr;
|
|
300
|
851 }
|
|
376
|
852 if (!spell_iswordp(mi.mi_end, wp->w_buffer))
|
|
344
|
853 return (int)(mi.mi_end - ptr);
|
|
346
|
854
|
|
|
855 /* Try including the digits in the word. */
|
|
|
856 mi.mi_fend = ptr + nrlen;
|
|
|
857 }
|
|
|
858 else
|
|
|
859 mi.mi_fend = ptr;
|
|
|
860
|
|
|
861 /* Find the normal end of the word (until the next non-word character). */
|
|
344
|
862 mi.mi_word = ptr;
|
|
376
|
863 if (spell_iswordp(mi.mi_fend, wp->w_buffer))
|
|
344
|
864 {
|
|
|
865 do
|
|
|
866 {
|
|
307
|
867 mb_ptr_adv(mi.mi_fend);
|
|
376
|
868 } while (*mi.mi_fend != NUL && spell_iswordp(mi.mi_fend, wp->w_buffer));
|
|
385
|
869
|
|
|
870 if (capcol != NULL && *capcol == 0 && wp->w_buffer->b_cap_prog != NULL)
|
|
|
871 {
|
|
|
872 /* Check word starting with capital letter. */
|
|
455
|
873 c = PTR2CHAR(ptr);
|
|
385
|
874 if (!SPELL_ISUPPER(c))
|
|
483
|
875 wrongcaplen = (int)(mi.mi_fend - ptr);
|
|
385
|
876 }
|
|
|
877 }
|
|
|
878 if (capcol != NULL)
|
|
|
879 *capcol = -1;
|
|
344
|
880
|
|
|
881 /* We always use the characters up to the next non-word character,
|
|
|
882 * also for bad words. */
|
|
|
883 mi.mi_end = mi.mi_fend;
|
|
|
884
|
|
|
885 /* Check caps type later. */
|
|
376
|
886 mi.mi_buf = wp->w_buffer;
|
|
344
|
887
|
|
483
|
888 /* case-fold the word with one non-word character, so that we can check
|
|
|
889 * for the word end. */
|
|
344
|
890 if (*mi.mi_fend != NUL)
|
|
|
891 mb_ptr_adv(mi.mi_fend);
|
|
|
892
|
|
|
893 (void)spell_casefold(ptr, (int)(mi.mi_fend - ptr), mi.mi_fword,
|
|
|
894 MAXWLEN + 1);
|
|
|
895 mi.mi_fwordlen = STRLEN(mi.mi_fword);
|
|
|
896
|
|
|
897 /* The word is bad unless we recognize it. */
|
|
|
898 mi.mi_result = SP_BAD;
|
|
492
|
899 mi.mi_result2 = SP_BAD;
|
|
344
|
900
|
|
|
901 /*
|
|
|
902 * Loop over the languages specified in 'spelllang'.
|
|
|
903 * We check them all, because a matching word may be longer than an
|
|
|
904 * already found matching word.
|
|
|
905 */
|
|
|
906 for (mi.mi_lp = LANGP_ENTRY(wp->w_buffer->b_langp, 0);
|
|
|
907 mi.mi_lp->lp_slang != NULL; ++mi.mi_lp)
|
|
|
908 {
|
|
|
909 /* Check for a matching word in case-folded words. */
|
|
|
910 find_word(&mi, FIND_FOLDWORD);
|
|
|
911
|
|
|
912 /* Check for a matching word in keep-case words. */
|
|
|
913 find_word(&mi, FIND_KEEPWORD);
|
|
|
914
|
|
|
915 /* Check for matching prefixes. */
|
|
485
|
916 find_prefix(&mi, FIND_FOLDWORD);
|
|
492
|
917
|
|
|
918 /* For a NOBREAK language, may want to use a word without a following
|
|
|
919 * word as a backup. */
|
|
|
920 if (mi.mi_lp->lp_slang->sl_nobreak && mi.mi_result == SP_BAD
|
|
|
921 && mi.mi_result2 != SP_BAD)
|
|
|
922 {
|
|
|
923 mi.mi_result = mi.mi_result2;
|
|
|
924 mi.mi_end = mi.mi_end2;
|
|
|
925 }
|
|
344
|
926 }
|
|
|
927
|
|
|
928 if (mi.mi_result != SP_OK)
|
|
|
929 {
|
|
346
|
930 /* If we found a number skip over it. Allows for "42nd". Do flag
|
|
|
931 * rare and local words, e.g., "3GPP". */
|
|
344
|
932 if (nrlen > 0)
|
|
346
|
933 {
|
|
|
934 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED)
|
|
|
935 return nrlen;
|
|
|
936 }
|
|
344
|
937
|
|
|
938 /* When we are at a non-word character there is no error, just
|
|
|
939 * skip over the character (try looking for a word after it). */
|
|
390
|
940 else if (!spell_iswordp_nmw(ptr))
|
|
243
|
941 {
|
|
385
|
942 if (capcol != NULL && wp->w_buffer->b_cap_prog != NULL)
|
|
|
943 {
|
|
|
944 regmatch_T regmatch;
|
|
|
945
|
|
|
946 /* Check for end of sentence. */
|
|
|
947 regmatch.regprog = wp->w_buffer->b_cap_prog;
|
|
|
948 regmatch.rm_ic = FALSE;
|
|
|
949 if (vim_regexec(®match, ptr, 0))
|
|
|
950 *capcol = (int)(regmatch.endp[0] - ptr);
|
|
|
951 }
|
|
|
952
|
|
344
|
953 #ifdef FEAT_MBYTE
|
|
|
954 if (has_mbyte)
|
|
474
|
955 return (*mb_ptr2len)(ptr);
|
|
344
|
956 #endif
|
|
|
957 return 1;
|
|
300
|
958 }
|
|
483
|
959 else if (mi.mi_end == ptr)
|
|
|
960 /* Always include at least one character. Required for when there
|
|
|
961 * is a mixup in "midword". */
|
|
|
962 mb_ptr_adv(mi.mi_end);
|
|
492
|
963 else if (mi.mi_result == SP_BAD
|
|
|
964 && LANGP_ENTRY(wp->w_buffer->b_langp, 0)->lp_slang->sl_nobreak)
|
|
|
965 {
|
|
|
966 char_u *p, *fp;
|
|
|
967 int save_result = mi.mi_result;
|
|
|
968
|
|
|
969 /* First language in 'spelllang' is NOBREAK. Find first position
|
|
|
970 * at which any word would be valid. */
|
|
|
971 mi.mi_lp = LANGP_ENTRY(wp->w_buffer->b_langp, 0);
|
|
|
972 p = mi.mi_word;
|
|
|
973 fp = mi.mi_fword;
|
|
|
974 for (;;)
|
|
|
975 {
|
|
|
976 mb_ptr_adv(p);
|
|
|
977 mb_ptr_adv(fp);
|
|
|
978 if (p >= mi.mi_end)
|
|
|
979 break;
|
|
|
980 mi.mi_compoff = fp - mi.mi_fword;
|
|
|
981 find_word(&mi, FIND_COMPOUND);
|
|
|
982 if (mi.mi_result != SP_BAD)
|
|
|
983 {
|
|
|
984 mi.mi_end = p;
|
|
|
985 break;
|
|
|
986 }
|
|
|
987 }
|
|
|
988 mi.mi_result = save_result;
|
|
|
989 }
|
|
243
|
990
|
|
344
|
991 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED)
|
|
|
992 *attrp = highlight_attr[HLF_SPB];
|
|
|
993 else if (mi.mi_result == SP_RARE)
|
|
|
994 *attrp = highlight_attr[HLF_SPR];
|
|
|
995 else
|
|
|
996 *attrp = highlight_attr[HLF_SPL];
|
|
243
|
997 }
|
|
|
998
|
|
483
|
999 if (wrongcaplen > 0 && (mi.mi_result == SP_OK || mi.mi_result == SP_RARE))
|
|
|
1000 {
|
|
|
1001 /* Report SpellCap only when the word isn't badly spelled. */
|
|
|
1002 *attrp = highlight_attr[HLF_SPC];
|
|
|
1003 return wrongcaplen;
|
|
|
1004 }
|
|
|
1005
|
|
300
|
1006 return (int)(mi.mi_end - ptr);
|
|
236
|
1007 }
|
|
|
1008
|
|
|
1009 /*
|
|
300
|
1010 * Check if the word at "mip->mi_word" is in the tree.
|
|
339
|
1011 * When "mode" is FIND_FOLDWORD check in fold-case word tree.
|
|
|
1012 * When "mode" is FIND_KEEPWORD check in keep-case word tree.
|
|
|
1013 * When "mode" is FIND_PREFIX check for word after prefix in fold-case word
|
|
|
1014 * tree.
|
|
300
|
1015 *
|
|
|
1016 * For a match mip->mi_result is updated.
|
|
243
|
1017 */
|
|
|
1018 static void
|
|
339
|
1019 find_word(mip, mode)
|
|
243
|
1020 matchinf_T *mip;
|
|
339
|
1021 int mode;
|
|
243
|
1022 {
|
|
324
|
1023 idx_T arridx = 0;
|
|
300
|
1024 int endlen[MAXWLEN]; /* length at possible word endings */
|
|
324
|
1025 idx_T endidx[MAXWLEN]; /* possible word endings */
|
|
300
|
1026 int endidxcnt = 0;
|
|
|
1027 int len;
|
|
|
1028 int wlen = 0;
|
|
|
1029 int flen;
|
|
|
1030 int c;
|
|
|
1031 char_u *ptr;
|
|
324
|
1032 idx_T lo, hi, m;
|
|
243
|
1033 #ifdef FEAT_MBYTE
|
|
300
|
1034 char_u *s;
|
|
490
|
1035 #endif
|
|
339
|
1036 char_u *p;
|
|
307
|
1037 int res = SP_BAD;
|
|
300
|
1038 slang_T *slang = mip->mi_lp->lp_slang;
|
|
|
1039 unsigned flags;
|
|
|
1040 char_u *byts;
|
|
324
|
1041 idx_T *idxs;
|
|
481
|
1042 int word_ends;
|
|
485
|
1043 int prefix_found;
|
|
492
|
1044 int nobreak_result;
|
|
481
|
1045
|
|
|
1046 if (mode == FIND_KEEPWORD || mode == FIND_KEEPCOMPOUND)
|
|
236
|
1047 {
|
|
300
|
1048 /* Check for word with matching case in keep-case tree. */
|
|
|
1049 ptr = mip->mi_word;
|
|
|
1050 flen = 9999; /* no case folding, always enough bytes */
|
|
|
1051 byts = slang->sl_kbyts;
|
|
|
1052 idxs = slang->sl_kidxs;
|
|
481
|
1053
|
|
|
1054 if (mode == FIND_KEEPCOMPOUND)
|
|
|
1055 /* Skip over the previously found word(s). */
|
|
|
1056 wlen += mip->mi_compoff;
|
|
236
|
1057 }
|
|
|
1058 else
|
|
|
1059 {
|
|
300
|
1060 /* Check for case-folded in case-folded tree. */
|
|
|
1061 ptr = mip->mi_fword;
|
|
|
1062 flen = mip->mi_fwordlen; /* available case-folded bytes */
|
|
|
1063 byts = slang->sl_fbyts;
|
|
|
1064 idxs = slang->sl_fidxs;
|
|
339
|
1065
|
|
|
1066 if (mode == FIND_PREFIX)
|
|
|
1067 {
|
|
|
1068 /* Skip over the prefix. */
|
|
|
1069 wlen = mip->mi_prefixlen;
|
|
|
1070 flen -= mip->mi_prefixlen;
|
|
|
1071 }
|
|
481
|
1072 else if (mode == FIND_COMPOUND)
|
|
|
1073 {
|
|
|
1074 /* Skip over the previously found word(s). */
|
|
|
1075 wlen = mip->mi_compoff;
|
|
|
1076 flen -= mip->mi_compoff;
|
|
|
1077 }
|
|
|
1078
|
|
243
|
1079 }
|
|
|
1080
|
|
300
|
1081 if (byts == NULL)
|
|
|
1082 return; /* array is empty */
|
|
236
|
1083
|
|
|
1084 /*
|
|
307
|
1085 * Repeat advancing in the tree until:
|
|
|
1086 * - there is a byte that doesn't match,
|
|
|
1087 * - we reach the end of the tree,
|
|
|
1088 * - or we reach the end of the line.
|
|
236
|
1089 */
|
|
300
|
1090 for (;;)
|
|
236
|
1091 {
|
|
346
|
1092 if (flen <= 0 && *mip->mi_fend != NUL)
|
|
339
|
1093 flen = fold_more(mip);
|
|
300
|
1094
|
|
|
1095 len = byts[arridx++];
|
|
|
1096
|
|
|
1097 /* If the first possible byte is a zero the word could end here.
|
|
|
1098 * Remember this index, we first check for the longest word. */
|
|
|
1099 if (byts[arridx] == 0)
|
|
|
1100 {
|
|
307
|
1101 if (endidxcnt == MAXWLEN)
|
|
|
1102 {
|
|
|
1103 /* Must be a corrupted spell file. */
|
|
|
1104 EMSG(_(e_format));
|
|
|
1105 return;
|
|
|
1106 }
|
|
300
|
1107 endlen[endidxcnt] = wlen;
|
|
|
1108 endidx[endidxcnt++] = arridx++;
|
|
|
1109 --len;
|
|
|
1110
|
|
|
1111 /* Skip over the zeros, there can be several flag/region
|
|
|
1112 * combinations. */
|
|
|
1113 while (len > 0 && byts[arridx] == 0)
|
|
|
1114 {
|
|
|
1115 ++arridx;
|
|
|
1116 --len;
|
|
|
1117 }
|
|
|
1118 if (len == 0)
|
|
|
1119 break; /* no children, word must end here */
|
|
|
1120 }
|
|
|
1121
|
|
|
1122 /* Stop looking at end of the line. */
|
|
|
1123 if (ptr[wlen] == NUL)
|
|
|
1124 break;
|
|
|
1125
|
|
|
1126 /* Perform a binary search in the list of accepted bytes. */
|
|
|
1127 c = ptr[wlen];
|
|
346
|
1128 if (c == TAB) /* <Tab> is handled like <Space> */
|
|
|
1129 c = ' ';
|
|
300
|
1130 lo = arridx;
|
|
|
1131 hi = arridx + len - 1;
|
|
|
1132 while (lo < hi)
|
|
|
1133 {
|
|
|
1134 m = (lo + hi) / 2;
|
|
|
1135 if (byts[m] > c)
|
|
|
1136 hi = m - 1;
|
|
|
1137 else if (byts[m] < c)
|
|
|
1138 lo = m + 1;
|
|
|
1139 else
|
|
|
1140 {
|
|
|
1141 lo = hi = m;
|
|
|
1142 break;
|
|
236
|
1143 }
|
|
|
1144 }
|
|
300
|
1145
|
|
|
1146 /* Stop if there is no matching byte. */
|
|
|
1147 if (hi < lo || byts[lo] != c)
|
|
|
1148 break;
|
|
|
1149
|
|
|
1150 /* Continue at the child (if there is one). */
|
|
|
1151 arridx = idxs[lo];
|
|
|
1152 ++wlen;
|
|
|
1153 --flen;
|
|
346
|
1154
|
|
|
1155 /* One space in the good word may stand for several spaces in the
|
|
|
1156 * checked word. */
|
|
|
1157 if (c == ' ')
|
|
|
1158 {
|
|
|
1159 for (;;)
|
|
|
1160 {
|
|
|
1161 if (flen <= 0 && *mip->mi_fend != NUL)
|
|
|
1162 flen = fold_more(mip);
|
|
|
1163 if (ptr[wlen] != ' ' && ptr[wlen] != TAB)
|
|
|
1164 break;
|
|
|
1165 ++wlen;
|
|
|
1166 --flen;
|
|
|
1167 }
|
|
|
1168 }
|
|
236
|
1169 }
|
|
|
1170
|
|
300
|
1171 /*
|
|
|
1172 * Verify that one of the possible endings is valid. Try the longest
|
|
|
1173 * first.
|
|
|
1174 */
|
|
|
1175 while (endidxcnt > 0)
|
|
|
1176 {
|
|
|
1177 --endidxcnt;
|
|
|
1178 arridx = endidx[endidxcnt];
|
|
|
1179 wlen = endlen[endidxcnt];
|
|
236
|
1180
|
|
300
|
1181 #ifdef FEAT_MBYTE
|
|
|
1182 if ((*mb_head_off)(ptr, ptr + wlen) > 0)
|
|
|
1183 continue; /* not at first byte of character */
|
|
|
1184 #endif
|
|
376
|
1185 if (spell_iswordp(ptr + wlen, mip->mi_buf))
|
|
481
|
1186 {
|
|
492
|
1187 if (slang->sl_compprog == NULL && !slang->sl_nobreak)
|
|
481
|
1188 continue; /* next char is a word character */
|
|
|
1189 word_ends = FALSE;
|
|
|
1190 }
|
|
|
1191 else
|
|
|
1192 word_ends = TRUE;
|
|
485
|
1193 /* The prefix flag is before compound flags. Once a valid prefix flag
|
|
|
1194 * has been found we try compound flags. */
|
|
|
1195 prefix_found = FALSE;
|
|
300
|
1196
|
|
|
1197 #ifdef FEAT_MBYTE
|
|
339
|
1198 if (mode != FIND_KEEPWORD && has_mbyte)
|
|
300
|
1199 {
|
|
|
1200 /* Compute byte length in original word, length may change
|
|
339
|
1201 * when folding case. This can be slow, take a shortcut when the
|
|
|
1202 * case-folded word is equal to the keep-case word. */
|
|
300
|
1203 p = mip->mi_word;
|
|
339
|
1204 if (STRNCMP(ptr, p, wlen) != 0)
|
|
|
1205 {
|
|
|
1206 for (s = ptr; s < ptr + wlen; mb_ptr_adv(s))
|
|
|
1207 mb_ptr_adv(p);
|
|
|
1208 wlen = p - mip->mi_word;
|
|
|
1209 }
|
|
300
|
1210 }
|
|
|
1211 #endif
|
|
236
|
1212
|
|
339
|
1213 /* Check flags and region. For FIND_PREFIX check the condition and
|
|
|
1214 * prefix ID.
|
|
|
1215 * Repeat this if there are more flags/region alternatives until there
|
|
|
1216 * is a match. */
|
|
|
1217 res = SP_BAD;
|
|
|
1218 for (len = byts[arridx - 1]; len > 0 && byts[arridx] == 0;
|
|
|
1219 --len, ++arridx)
|
|
300
|
1220 {
|
|
|
1221 flags = idxs[arridx];
|
|
324
|
1222
|
|
339
|
1223 /* For the fold-case tree check that the case of the checked word
|
|
|
1224 * matches with what the word in the tree requires.
|
|
|
1225 * For keep-case tree the case is always right. For prefixes we
|
|
|
1226 * don't bother to check. */
|
|
|
1227 if (mode == FIND_FOLDWORD)
|
|
300
|
1228 {
|
|
|
1229 if (mip->mi_cend != mip->mi_word + wlen)
|
|
|
1230 {
|
|
323
|
1231 /* mi_capflags was set for a different word length, need
|
|
|
1232 * to do it again. */
|
|
300
|
1233 mip->mi_cend = mip->mi_word + wlen;
|
|
323
|
1234 mip->mi_capflags = captype(mip->mi_word, mip->mi_cend);
|
|
300
|
1235 }
|
|
|
1236
|
|
346
|
1237 if (mip->mi_capflags == WF_KEEPCAP
|
|
|
1238 || !spell_valid_case(mip->mi_capflags, flags))
|
|
339
|
1239 continue;
|
|
300
|
1240 }
|
|
236
|
1241
|
|
339
|
1242 /* When mode is FIND_PREFIX the word must support the prefix:
|
|
|
1243 * check the prefix ID and the condition. Do that for the list at
|
|
366
|
1244 * mip->mi_prefarridx that find_prefix() filled. */
|
|
485
|
1245 else if (mode == FIND_PREFIX && !prefix_found)
|
|
481
|
1246 {
|
|
366
|
1247 c = valid_word_prefix(mip->mi_prefcnt, mip->mi_prefarridx,
|
|
390
|
1248 flags,
|
|
455
|
1249 mip->mi_word + mip->mi_cprefixlen, slang,
|
|
|
1250 FALSE);
|
|
366
|
1251 if (c == 0)
|
|
339
|
1252 continue;
|
|
366
|
1253
|
|
|
1254 /* Use the WF_RARE flag for a rare prefix. */
|
|
|
1255 if (c & WF_RAREPFX)
|
|
|
1256 flags |= WF_RARE;
|
|
485
|
1257 prefix_found = TRUE;
|
|
339
|
1258 }
|
|
|
1259
|
|
492
|
1260 if (slang->sl_nobreak)
|
|
|
1261 {
|
|
|
1262 if ((mode == FIND_COMPOUND || mode == FIND_KEEPCOMPOUND)
|
|
|
1263 && (flags & WF_BANNED) == 0)
|
|
|
1264 {
|
|
|
1265 /* NOBREAK: found a valid following word. That's all we
|
|
|
1266 * need to know, so return. */
|
|
|
1267 mip->mi_result = SP_OK;
|
|
|
1268 break;
|
|
|
1269 }
|
|
|
1270 }
|
|
|
1271
|
|
|
1272 else if ((mode == FIND_COMPOUND || mode == FIND_KEEPCOMPOUND
|
|
|
1273 || !word_ends))
|
|
481
|
1274 {
|
|
483
|
1275 /* If there is no flag or the word is shorter than
|
|
|
1276 * COMPOUNDMIN reject it quickly.
|
|
|
1277 * Makes you wonder why someone puts a compound flag on a word
|
|
481
|
1278 * that's too short... Myspell compatibility requires this
|
|
|
1279 * anyway. */
|
|
490
|
1280 if (((unsigned)flags >> 24) == 0
|
|
|
1281 || wlen - mip->mi_compoff < slang->sl_compminlen)
|
|
483
|
1282 continue;
|
|
|
1283
|
|
490
|
1284 /* Limit the number of compound words to COMPOUNDMAX if no
|
|
|
1285 * maximum for syllables is specified. */
|
|
|
1286 if (!word_ends && mip->mi_complen + 2 > slang->sl_compmax
|
|
|
1287 && slang->sl_compsylmax == MAXWLEN)
|
|
483
|
1288 continue;
|
|
|
1289
|
|
485
|
1290 /* Quickly check if compounding is possible with this flag. */
|
|
|
1291 if (vim_strchr(mip->mi_complen == 0
|
|
|
1292 ? slang->sl_compstartflags
|
|
|
1293 : slang->sl_compallflags,
|
|
483
|
1294 ((unsigned)flags >> 24)) == NULL)
|
|
481
|
1295 continue;
|
|
|
1296
|
|
490
|
1297 if (mode == FIND_COMPOUND)
|
|
|
1298 {
|
|
|
1299 int capflags;
|
|
|
1300
|
|
|
1301 /* Need to check the caps type of the appended compound
|
|
|
1302 * word. */
|
|
|
1303 #ifdef FEAT_MBYTE
|
|
|
1304 if (has_mbyte && STRNCMP(ptr, mip->mi_word,
|
|
|
1305 mip->mi_compoff) != 0)
|
|
|
1306 {
|
|
|
1307 /* case folding may have changed the length */
|
|
|
1308 p = mip->mi_word;
|
|
|
1309 for (s = ptr; s < ptr + mip->mi_compoff; mb_ptr_adv(s))
|
|
|
1310 mb_ptr_adv(p);
|
|
|
1311 }
|
|
|
1312 else
|
|
|
1313 #endif
|
|
|
1314 p = mip->mi_word + mip->mi_compoff;
|
|
|
1315 capflags = captype(p, mip->mi_word + wlen);
|
|
|
1316 if (capflags == WF_KEEPCAP || (capflags == WF_ALLCAP
|
|
|
1317 && (flags & WF_FIXCAP) != 0))
|
|
|
1318 continue;
|
|
|
1319
|
|
|
1320 if (capflags != WF_ALLCAP)
|
|
|
1321 {
|
|
|
1322 /* When the character before the word is a word
|
|
|
1323 * character we do not accept a Onecap word. We do
|
|
|
1324 * accept a no-caps word, even when the dictionary
|
|
|
1325 * word specifies ONECAP. */
|
|
|
1326 mb_ptr_back(mip->mi_word, p);
|
|
|
1327 if (spell_iswordp_nmw(p)
|
|
|
1328 ? capflags == WF_ONECAP
|
|
|
1329 : (flags & WF_ONECAP) != 0
|
|
|
1330 && capflags != WF_ONECAP)
|
|
|
1331 continue;
|
|
|
1332 }
|
|
|
1333 }
|
|
|
1334
|
|
483
|
1335 /* If the word ends the sequence of compound flags of the
|
|
|
1336 * words must match with one of the COMPOUNDFLAGS items and
|
|
|
1337 * the number of syllables must not be too large. */
|
|
|
1338 mip->mi_compflags[mip->mi_complen] = ((unsigned)flags >> 24);
|
|
|
1339 mip->mi_compflags[mip->mi_complen + 1] = NUL;
|
|
|
1340 if (word_ends)
|
|
|
1341 {
|
|
|
1342 char_u fword[MAXWLEN];
|
|
|
1343
|
|
|
1344 if (slang->sl_compsylmax < MAXWLEN)
|
|
|
1345 {
|
|
|
1346 /* "fword" is only needed for checking syllables. */
|
|
|
1347 if (ptr == mip->mi_word)
|
|
|
1348 (void)spell_casefold(ptr, wlen, fword, MAXWLEN);
|
|
|
1349 else
|
|
|
1350 vim_strncpy(fword, ptr, endlen[endidxcnt]);
|
|
|
1351 }
|
|
|
1352 if (!can_compound(slang, fword, mip->mi_compflags))
|
|
|
1353 continue;
|
|
|
1354 }
|
|
481
|
1355 }
|
|
|
1356
|
|
492
|
1357 nobreak_result = SP_OK;
|
|
|
1358
|
|
481
|
1359 if (!word_ends)
|
|
|
1360 {
|
|
492
|
1361 int save_result = mip->mi_result;
|
|
|
1362 char_u *save_end = mip->mi_end;
|
|
|
1363
|
|
|
1364 /* Check that a valid word follows. If there is one and we
|
|
|
1365 * are compounding, it will set "mi_result", thus we are
|
|
|
1366 * always finished here. For NOBREAK we only check that a
|
|
|
1367 * valid word follows.
|
|
481
|
1368 * Recursive! */
|
|
492
|
1369 if (slang->sl_nobreak)
|
|
|
1370 mip->mi_result = SP_BAD;
|
|
481
|
1371
|
|
|
1372 /* Find following word in case-folded tree. */
|
|
|
1373 mip->mi_compoff = endlen[endidxcnt];
|
|
|
1374 #ifdef FEAT_MBYTE
|
|
|
1375 if (has_mbyte && mode == FIND_KEEPWORD)
|
|
|
1376 {
|
|
|
1377 /* Compute byte length in case-folded word from "wlen":
|
|
|
1378 * byte length in keep-case word. Length may change when
|
|
|
1379 * folding case. This can be slow, take a shortcut when
|
|
|
1380 * the case-folded word is equal to the keep-case word. */
|
|
|
1381 p = mip->mi_fword;
|
|
|
1382 if (STRNCMP(ptr, p, wlen) != 0)
|
|
|
1383 {
|
|
|
1384 for (s = ptr; s < ptr + wlen; mb_ptr_adv(s))
|
|
|
1385 mb_ptr_adv(p);
|
|
|
1386 mip->mi_compoff = p - mip->mi_fword;
|
|
|
1387 }
|
|
|
1388 }
|
|
|
1389 #endif
|
|
485
|
1390 c = mip->mi_compoff;
|
|
483
|
1391 ++mip->mi_complen;
|
|
481
|
1392 find_word(mip, FIND_COMPOUND);
|
|
|
1393
|
|
492
|
1394 /* When NOBREAK any word that matches is OK. Otherwise we
|
|
|
1395 * need to find the longest match, thus try with keep-case and
|
|
|
1396 * prefix too. */
|
|
|
1397 if (!slang->sl_nobreak || mip->mi_result == SP_BAD)
|
|
|
1398 {
|
|
|
1399 /* Find following word in keep-case tree. */
|
|
|
1400 mip->mi_compoff = wlen;
|
|
|
1401 find_word(mip, FIND_KEEPCOMPOUND);
|
|
|
1402
|
|
|
1403 if (!slang->sl_nobreak || mip->mi_result == SP_BAD)
|
|
|
1404 {
|
|
|
1405 /* Check for following word with prefix. */
|
|
|
1406 mip->mi_compoff = c;
|
|
|
1407 find_prefix(mip, FIND_COMPOUND);
|
|
|
1408 }
|
|
|
1409 }
|
|
483
|
1410 --mip->mi_complen;
|
|
485
|
1411
|
|
492
|
1412 if (slang->sl_nobreak)
|
|
|
1413 {
|
|
|
1414 nobreak_result = mip->mi_result;
|
|
|
1415 mip->mi_result = save_result;
|
|
|
1416 mip->mi_end = save_end;
|
|
|
1417 }
|
|
|
1418 else
|
|
|
1419 {
|
|
|
1420 if (mip->mi_result == SP_OK)
|
|
|
1421 break;
|
|
|
1422 continue;
|
|
|
1423 }
|
|
481
|
1424 }
|
|
|
1425
|
|
339
|
1426 if (flags & WF_BANNED)
|
|
|
1427 res = SP_BANNED;
|
|
|
1428 else if (flags & WF_REGION)
|
|
|
1429 {
|
|
|
1430 /* Check region. */
|
|
390
|
1431 if ((mip->mi_lp->lp_region & (flags >> 16)) != 0)
|
|
300
|
1432 res = SP_OK;
|
|
339
|
1433 else
|
|
|
1434 res = SP_LOCAL;
|
|
300
|
1435 }
|
|
339
|
1436 else if (flags & WF_RARE)
|
|
|
1437 res = SP_RARE;
|
|
307
|
1438 else
|
|
339
|
1439 res = SP_OK;
|
|
|
1440
|
|
492
|
1441 /* Always use the longest match and the best result. For NOBREAK
|
|
|
1442 * we separately keep the longest match without a following good
|
|
|
1443 * word as a fall-back. */
|
|
|
1444 if (nobreak_result == SP_BAD)
|
|
|
1445 {
|
|
|
1446 if (mip->mi_result2 > res)
|
|
|
1447 {
|
|
|
1448 mip->mi_result2 = res;
|
|
|
1449 mip->mi_end2 = mip->mi_word + wlen;
|
|
|
1450 }
|
|
|
1451 else if (mip->mi_result2 == res
|
|
|
1452 && mip->mi_end2 < mip->mi_word + wlen)
|
|
|
1453 mip->mi_end2 = mip->mi_word + wlen;
|
|
|
1454 }
|
|
|
1455 else if (mip->mi_result > res)
|
|
339
|
1456 {
|
|
|
1457 mip->mi_result = res;
|
|
|
1458 mip->mi_end = mip->mi_word + wlen;
|
|
|
1459 }
|
|
351
|
1460 else if (mip->mi_result == res && mip->mi_end < mip->mi_word + wlen)
|
|
339
|
1461 mip->mi_end = mip->mi_word + wlen;
|
|
|
1462
|
|
492
|
1463 if (mip->mi_result == SP_OK)
|
|
339
|
1464 break;
|
|
300
|
1465 }
|
|
|
1466
|
|
492
|
1467 if (mip->mi_result == SP_OK)
|
|
300
|
1468 break;
|
|
|
1469 }
|
|
236
|
1470 }
|
|
|
1471
|
|
323
|
1472 /*
|
|
483
|
1473 * Return TRUE if "flags" is a valid sequence of compound flags and
|
|
|
1474 * "word[len]" does not have too many syllables.
|
|
482
|
1475 */
|
|
|
1476 static int
|
|
483
|
1477 can_compound(slang, word, flags)
|
|
482
|
1478 slang_T *slang;
|
|
483
|
1479 char_u *word;
|
|
|
1480 char_u *flags;
|
|
|
1481 {
|
|
|
1482 regmatch_T regmatch;
|
|
|
1483
|
|
|
1484 if (slang->sl_compprog == NULL)
|
|
|
1485 return FALSE;
|
|
|
1486 regmatch.regprog = slang->sl_compprog;
|
|
|
1487 regmatch.rm_ic = FALSE;
|
|
|
1488 if (!vim_regexec(®match, flags, 0))
|
|
|
1489 return FALSE;
|
|
|
1490
|
|
490
|
1491 /* Count the number of syllables. This may be slow, do it last. If there
|
|
|
1492 * are too many syllables AND the number of compound words is above
|
|
|
1493 * COMPOUNDMAX then compounding is not allowed. */
|
|
483
|
1494 if (slang->sl_compsylmax < MAXWLEN
|
|
|
1495 && count_syllables(slang, word) > slang->sl_compsylmax)
|
|
490
|
1496 return STRLEN(flags) < slang->sl_compmax;
|
|
483
|
1497 return TRUE;
|
|
482
|
1498 }
|
|
|
1499
|
|
|
1500 /*
|
|
390
|
1501 * Return non-zero if the prefix indicated by "arridx" matches with the prefix
|
|
|
1502 * ID in "flags" for the word "word".
|
|
366
|
1503 * The WF_RAREPFX flag is included in the return value for a rare prefix.
|
|
351
|
1504 */
|
|
|
1505 static int
|
|
455
|
1506 valid_word_prefix(totprefcnt, arridx, flags, word, slang, cond_req)
|
|
351
|
1507 int totprefcnt; /* nr of prefix IDs */
|
|
|
1508 int arridx; /* idx in sl_pidxs[] */
|
|
390
|
1509 int flags;
|
|
351
|
1510 char_u *word;
|
|
|
1511 slang_T *slang;
|
|
455
|
1512 int cond_req; /* only use prefixes with a condition */
|
|
351
|
1513 {
|
|
|
1514 int prefcnt;
|
|
|
1515 int pidx;
|
|
|
1516 regprog_T *rp;
|
|
|
1517 regmatch_T regmatch;
|
|
390
|
1518 int prefid;
|
|
|
1519
|
|
|
1520 prefid = (unsigned)flags >> 24;
|
|
351
|
1521 for (prefcnt = totprefcnt - 1; prefcnt >= 0; --prefcnt)
|
|
|
1522 {
|
|
|
1523 pidx = slang->sl_pidxs[arridx + prefcnt];
|
|
|
1524
|
|
|
1525 /* Check the prefix ID. */
|
|
|
1526 if (prefid != (pidx & 0xff))
|
|
|
1527 continue;
|
|
|
1528
|
|
390
|
1529 /* Check if the prefix doesn't combine and the word already has a
|
|
|
1530 * suffix. */
|
|
|
1531 if ((flags & WF_HAS_AFF) && (pidx & WF_PFX_NC))
|
|
|
1532 continue;
|
|
|
1533
|
|
351
|
1534 /* Check the condition, if there is one. The condition index is
|
|
366
|
1535 * stored in the two bytes above the prefix ID byte. */
|
|
|
1536 rp = slang->sl_prefprog[((unsigned)pidx >> 8) & 0xffff];
|
|
351
|
1537 if (rp != NULL)
|
|
|
1538 {
|
|
|
1539 regmatch.regprog = rp;
|
|
|
1540 regmatch.rm_ic = FALSE;
|
|
|
1541 if (!vim_regexec(®match, word, 0))
|
|
|
1542 continue;
|
|
|
1543 }
|
|
455
|
1544 else if (cond_req)
|
|
|
1545 continue;
|
|
|
1546
|
|
|
1547 /* It's a match! Return the WF_ flags. */
|
|
366
|
1548 return pidx;
|
|
|
1549 }
|
|
|
1550 return 0;
|
|
351
|
1551 }
|
|
|
1552
|
|
|
1553 /*
|
|
339
|
1554 * Check if the word at "mip->mi_word" has a matching prefix.
|
|
|
1555 * If it does, then check the following word.
|
|
|
1556 *
|
|
485
|
1557 * If "mode" is "FIND_COMPOUND" then do the same after another word, find a
|
|
|
1558 * prefix in a compound word.
|
|
|
1559 *
|
|
339
|
1560 * For a match mip->mi_result is updated.
|
|
|
1561 */
|
|
|
1562 static void
|
|
485
|
1563 find_prefix(mip, mode)
|
|
339
|
1564 matchinf_T *mip;
|
|
485
|
1565 int mode;
|
|
339
|
1566 {
|
|
|
1567 idx_T arridx = 0;
|
|
|
1568 int len;
|
|
|
1569 int wlen = 0;
|
|
|
1570 int flen;
|
|
|
1571 int c;
|
|
|
1572 char_u *ptr;
|
|
|
1573 idx_T lo, hi, m;
|
|
|
1574 slang_T *slang = mip->mi_lp->lp_slang;
|
|
|
1575 char_u *byts;
|
|
|
1576 idx_T *idxs;
|
|
|
1577
|
|
375
|
1578 byts = slang->sl_pbyts;
|
|
|
1579 if (byts == NULL)
|
|
|
1580 return; /* array is empty */
|
|
|
1581
|
|
339
|
1582 /* We use the case-folded word here, since prefixes are always
|
|
|
1583 * case-folded. */
|
|
|
1584 ptr = mip->mi_fword;
|
|
|
1585 flen = mip->mi_fwordlen; /* available case-folded bytes */
|
|
485
|
1586 if (mode == FIND_COMPOUND)
|
|
|
1587 {
|
|
|
1588 /* Skip over the previously found word(s). */
|
|
|
1589 ptr += mip->mi_compoff;
|
|
|
1590 flen -= mip->mi_compoff;
|
|
|
1591 }
|
|
339
|
1592 idxs = slang->sl_pidxs;
|
|
|
1593
|
|
|
1594 /*
|
|
|
1595 * Repeat advancing in the tree until:
|
|
|
1596 * - there is a byte that doesn't match,
|
|
|
1597 * - we reach the end of the tree,
|
|
|
1598 * - or we reach the end of the line.
|
|
|
1599 */
|
|
|
1600 for (;;)
|
|
|
1601 {
|
|
|
1602 if (flen == 0 && *mip->mi_fend != NUL)
|
|
|
1603 flen = fold_more(mip);
|
|
|
1604
|
|
|
1605 len = byts[arridx++];
|
|
|
1606
|
|
|
1607 /* If the first possible byte is a zero the prefix could end here.
|
|
|
1608 * Check if the following word matches and supports the prefix. */
|
|
|
1609 if (byts[arridx] == 0)
|
|
|
1610 {
|
|
|
1611 /* There can be several prefixes with different conditions. We
|
|
|
1612 * try them all, since we don't know which one will give the
|
|
|
1613 * longest match. The word is the same each time, pass the list
|
|
|
1614 * of possible prefixes to find_word(). */
|
|
|
1615 mip->mi_prefarridx = arridx;
|
|
|
1616 mip->mi_prefcnt = len;
|
|
|
1617 while (len > 0 && byts[arridx] == 0)
|
|
|
1618 {
|
|
|
1619 ++arridx;
|
|
|
1620 --len;
|
|
|
1621 }
|
|
|
1622 mip->mi_prefcnt -= len;
|
|
|
1623
|
|
|
1624 /* Find the word that comes after the prefix. */
|
|
|
1625 mip->mi_prefixlen = wlen;
|
|
485
|
1626 if (mode == FIND_COMPOUND)
|
|
|
1627 /* Skip over the previously found word(s). */
|
|
|
1628 mip->mi_prefixlen += mip->mi_compoff;
|
|
|
1629
|
|
455
|
1630 #ifdef FEAT_MBYTE
|
|
|
1631 if (has_mbyte)
|
|
|
1632 {
|
|
|
1633 /* Case-folded length may differ from original length. */
|
|
485
|
1634 mip->mi_cprefixlen = nofold_len(mip->mi_fword,
|
|
|
1635 mip->mi_prefixlen, mip->mi_word);
|
|
455
|
1636 }
|
|
|
1637 else
|
|
485
|
1638 mip->mi_cprefixlen = mip->mi_prefixlen;
|
|
455
|
1639 #endif
|
|
339
|
1640 find_word(mip, FIND_PREFIX);
|
|
|
1641
|
|
|
1642
|
|
|
1643 if (len == 0)
|
|
|
1644 break; /* no children, word must end here */
|
|
|
1645 }
|
|
|
1646
|
|
|
1647 /* Stop looking at end of the line. */
|
|
|
1648 if (ptr[wlen] == NUL)
|
|
|
1649 break;
|
|
|
1650
|
|
|
1651 /* Perform a binary search in the list of accepted bytes. */
|
|
|
1652 c = ptr[wlen];
|
|
|
1653 lo = arridx;
|
|
|
1654 hi = arridx + len - 1;
|
|
|
1655 while (lo < hi)
|
|
|
1656 {
|
|
|
1657 m = (lo + hi) / 2;
|
|
|
1658 if (byts[m] > c)
|
|
|
1659 hi = m - 1;
|
|
|
1660 else if (byts[m] < c)
|
|
|
1661 lo = m + 1;
|
|
|
1662 else
|
|
|
1663 {
|
|
|
1664 lo = hi = m;
|
|
|
1665 break;
|
|
|
1666 }
|
|
|
1667 }
|
|
|
1668
|
|
|
1669 /* Stop if there is no matching byte. */
|
|
|
1670 if (hi < lo || byts[lo] != c)
|
|
|
1671 break;
|
|
|
1672
|
|
|
1673 /* Continue at the child (if there is one). */
|
|
|
1674 arridx = idxs[lo];
|
|
|
1675 ++wlen;
|
|
|
1676 --flen;
|
|
|
1677 }
|
|
|
1678 }
|
|
|
1679
|
|
|
1680 /*
|
|
|
1681 * Need to fold at least one more character. Do until next non-word character
|
|
|
1682 * for efficiency.
|
|
|
1683 * Return the length of the folded chars in bytes.
|
|
|
1684 */
|
|
|
1685 static int
|
|
|
1686 fold_more(mip)
|
|
|
1687 matchinf_T *mip;
|
|
|
1688 {
|
|
|
1689 int flen;
|
|
|
1690 char_u *p;
|
|
|
1691
|
|
|
1692 p = mip->mi_fend;
|
|
|
1693 do
|
|
|
1694 {
|
|
|
1695 mb_ptr_adv(mip->mi_fend);
|
|
376
|
1696 } while (*mip->mi_fend != NUL && spell_iswordp(mip->mi_fend, mip->mi_buf));
|
|
339
|
1697
|
|
|
1698 /* Include the non-word character so that we can check for the
|
|
|
1699 * word end. */
|
|
|
1700 if (*mip->mi_fend != NUL)
|
|
|
1701 mb_ptr_adv(mip->mi_fend);
|
|
|
1702
|
|
|
1703 (void)spell_casefold(p, (int)(mip->mi_fend - p),
|
|
|
1704 mip->mi_fword + mip->mi_fwordlen,
|
|
|
1705 MAXWLEN - mip->mi_fwordlen);
|
|
|
1706 flen = STRLEN(mip->mi_fword + mip->mi_fwordlen);
|
|
|
1707 mip->mi_fwordlen += flen;
|
|
|
1708 return flen;
|
|
|
1709 }
|
|
|
1710
|
|
|
1711 /*
|
|
323
|
1712 * Check case flags for a word. Return TRUE if the word has the requested
|
|
|
1713 * case.
|
|
|
1714 */
|
|
|
1715 static int
|
|
389
|
1716 spell_valid_case(wordflags, treeflags)
|
|
|
1717 int wordflags; /* flags for the checked word. */
|
|
323
|
1718 int treeflags; /* flags for the word in the spell tree */
|
|
|
1719 {
|
|
389
|
1720 return ((wordflags == WF_ALLCAP && (treeflags & WF_FIXCAP) == 0)
|
|
323
|
1721 || ((treeflags & (WF_ALLCAP | WF_KEEPCAP)) == 0
|
|
474
|
1722 && ((treeflags & WF_ONECAP) == 0
|
|
|
1723 || (wordflags & WF_ONECAP) != 0)));
|
|
323
|
1724 }
|
|
|
1725
|
|
351
|
1726 /*
|
|
|
1727 * Return TRUE if spell checking is not enabled.
|
|
|
1728 */
|
|
|
1729 static int
|
|
|
1730 no_spell_checking()
|
|
|
1731 {
|
|
|
1732 if (!curwin->w_p_spell || *curbuf->b_p_spl == NUL)
|
|
|
1733 {
|
|
|
1734 EMSG(_("E756: Spell checking is not enabled"));
|
|
|
1735 return TRUE;
|
|
|
1736 }
|
|
|
1737 return FALSE;
|
|
|
1738 }
|
|
300
|
1739
|
|
236
|
1740 /*
|
|
|
1741 * Move to next spell error.
|
|
323
|
1742 * "curline" is TRUE for "z?": find word under/after cursor in the same line.
|
|
483
|
1743 * For Insert mode completion "dir" is BACKWARD and "curline" is TRUE: move
|
|
|
1744 * to after badly spelled word before the cursor.
|
|
236
|
1745 * Return OK if found, FAIL otherwise.
|
|
|
1746 */
|
|
|
1747 int
|
|
323
|
1748 spell_move_to(dir, allwords, curline)
|
|
236
|
1749 int dir; /* FORWARD or BACKWARD */
|
|
|
1750 int allwords; /* TRUE for "[s" and "]s" */
|
|
323
|
1751 int curline;
|
|
236
|
1752 {
|
|
249
|
1753 linenr_T lnum;
|
|
|
1754 pos_T found_pos;
|
|
236
|
1755 char_u *line;
|
|
|
1756 char_u *p;
|
|
346
|
1757 char_u *endp;
|
|
|
1758 int attr;
|
|
236
|
1759 int len;
|
|
249
|
1760 int has_syntax = syntax_present(curbuf);
|
|
|
1761 int col;
|
|
|
1762 int can_spell;
|
|
346
|
1763 char_u *buf = NULL;
|
|
|
1764 int buflen = 0;
|
|
|
1765 int skip = 0;
|
|
385
|
1766 int capcol = -1;
|
|
236
|
1767
|
|
351
|
1768 if (no_spell_checking())
|
|
236
|
1769 return FAIL;
|
|
|
1770
|
|
249
|
1771 /*
|
|
|
1772 * Start looking for bad word at the start of the line, because we can't
|
|
346
|
1773 * start halfway a word, we don't know where the it starts or ends.
|
|
249
|
1774 *
|
|
|
1775 * When searching backwards, we continue in the line to find the last
|
|
|
1776 * bad word (in the cursor line: before the cursor).
|
|
346
|
1777 *
|
|
|
1778 * We concatenate the start of the next line, so that wrapped words work
|
|
|
1779 * (e.g. "et<line-break>cetera"). Doesn't work when searching backwards
|
|
|
1780 * though...
|
|
249
|
1781 */
|
|
|
1782 lnum = curwin->w_cursor.lnum;
|
|
|
1783 found_pos.lnum = 0;
|
|
236
|
1784
|
|
|
1785 while (!got_int)
|
|
|
1786 {
|
|
249
|
1787 line = ml_get(lnum);
|
|
346
|
1788
|
|
|
1789 len = STRLEN(line);
|
|
|
1790 if (buflen < len + MAXWLEN + 2)
|
|
|
1791 {
|
|
|
1792 vim_free(buf);
|
|
|
1793 buflen = len + MAXWLEN + 2;
|
|
|
1794 buf = alloc(buflen);
|
|
|
1795 if (buf == NULL)
|
|
|
1796 break;
|
|
|
1797 }
|
|
|
1798
|
|
385
|
1799 /* In first line check first word for Capital. */
|
|
|
1800 if (lnum == 1)
|
|
|
1801 capcol = 0;
|
|
|
1802
|
|
|
1803 /* For checking first word with a capital skip white space. */
|
|
|
1804 if (capcol == 0)
|
|
|
1805 capcol = skipwhite(line) - line;
|
|
|
1806
|
|
346
|
1807 /* Copy the line into "buf" and append the start of the next line if
|
|
|
1808 * possible. */
|
|
|
1809 STRCPY(buf, line);
|
|
|
1810 if (lnum < curbuf->b_ml.ml_line_count)
|
|
|
1811 spell_cat_line(buf + STRLEN(buf), ml_get(lnum + 1), MAXWLEN);
|
|
|
1812
|
|
|
1813 p = buf + skip;
|
|
|
1814 endp = buf + len;
|
|
|
1815 while (p < endp)
|
|
236
|
1816 {
|
|
300
|
1817 /* When searching backward don't search after the cursor. */
|
|
|
1818 if (dir == BACKWARD
|
|
|
1819 && lnum == curwin->w_cursor.lnum
|
|
346
|
1820 && (colnr_T)(p - buf) >= curwin->w_cursor.col)
|
|
300
|
1821 break;
|
|
249
|
1822
|
|
300
|
1823 /* start of word */
|
|
346
|
1824 attr = 0;
|
|
385
|
1825 len = spell_check(curwin, p, &attr, &capcol);
|
|
249
|
1826
|
|
300
|
1827 if (attr != 0)
|
|
|
1828 {
|
|
|
1829 /* We found a bad word. Check the attribute. */
|
|
|
1830 if (allwords || attr == highlight_attr[HLF_SPB])
|
|
236
|
1831 {
|
|
300
|
1832 /* When searching forward only accept a bad word after
|
|
|
1833 * the cursor. */
|
|
|
1834 if (dir == BACKWARD
|
|
|
1835 || lnum > curwin->w_cursor.lnum
|
|
|
1836 || (lnum == curwin->w_cursor.lnum
|
|
346
|
1837 && (colnr_T)(curline ? p - buf + len
|
|
|
1838 : p - buf)
|
|
300
|
1839 > curwin->w_cursor.col))
|
|
236
|
1840 {
|
|
300
|
1841 if (has_syntax)
|
|
249
|
1842 {
|
|
346
|
1843 col = p - buf;
|
|
300
|
1844 (void)syn_get_id(lnum, (colnr_T)col,
|
|
|
1845 FALSE, &can_spell);
|
|
|
1846 }
|
|
|
1847 else
|
|
|
1848 can_spell = TRUE;
|
|
249
|
1849
|
|
300
|
1850 if (can_spell)
|
|
|
1851 {
|
|
|
1852 found_pos.lnum = lnum;
|
|
346
|
1853 found_pos.col = p - buf;
|
|
249
|
1854 #ifdef FEAT_VIRTUALEDIT
|
|
300
|
1855 found_pos.coladd = 0;
|
|
249
|
1856 #endif
|
|
300
|
1857 if (dir == FORWARD)
|
|
|
1858 {
|
|
|
1859 /* No need to search further. */
|
|
|
1860 curwin->w_cursor = found_pos;
|
|
346
|
1861 vim_free(buf);
|
|
300
|
1862 return OK;
|
|
249
|
1863 }
|
|
483
|
1864 else if (curline)
|
|
|
1865 /* Insert mode completion: put cursor after
|
|
|
1866 * the bad word. */
|
|
|
1867 found_pos.col += len;
|
|
249
|
1868 }
|
|
236
|
1869 }
|
|
|
1870 }
|
|
|
1871 }
|
|
|
1872
|
|
300
|
1873 /* advance to character after the word */
|
|
|
1874 p += len;
|
|
385
|
1875 capcol -= len;
|
|
236
|
1876 }
|
|
|
1877
|
|
483
|
1878 if (dir == BACKWARD && found_pos.lnum != 0)
|
|
|
1879 {
|
|
|
1880 /* Use the last match in the line. */
|
|
|
1881 curwin->w_cursor = found_pos;
|
|
|
1882 vim_free(buf);
|
|
|
1883 return OK;
|
|
|
1884 }
|
|
|
1885
|
|
323
|
1886 if (curline)
|
|
346
|
1887 break; /* only check cursor line */
|
|
323
|
1888
|
|
236
|
1889 /* Advance to next line. */
|
|
249
|
1890 if (dir == BACKWARD)
|
|
|
1891 {
|
|
|
1892 if (lnum == 1)
|
|
346
|
1893 break;
|
|
249
|
1894 --lnum;
|
|
385
|
1895 capcol = -1;
|
|
249
|
1896 }
|
|
|
1897 else
|
|
|
1898 {
|
|
|
1899 if (lnum == curbuf->b_ml.ml_line_count)
|
|
346
|
1900 break;
|
|
249
|
1901 ++lnum;
|
|
346
|
1902
|
|
|
1903 /* Skip the characters at the start of the next line that were
|
|
|
1904 * included in a match crossing line boundaries. */
|
|
|
1905 if (attr == 0)
|
|
|
1906 skip = p - endp;
|
|
|
1907 else
|
|
|
1908 skip = 0;
|
|
385
|
1909
|
|
|
1910 /* Capscol skips over the inserted space. */
|
|
|
1911 --capcol;
|
|
|
1912
|
|
|
1913 /* But after empty line check first word in next line */
|
|
|
1914 if (*skipwhite(line) == NUL)
|
|
|
1915 capcol = 0;
|
|
249
|
1916 }
|
|
236
|
1917
|
|
|
1918 line_breakcheck();
|
|
|
1919 }
|
|
|
1920
|
|
346
|
1921 vim_free(buf);
|
|
|
1922 return FAIL;
|
|
|
1923 }
|
|
|
1924
|
|
|
1925 /*
|
|
|
1926 * For spell checking: concatenate the start of the following line "line" into
|
|
|
1927 * "buf", blanking-out special characters. Copy less then "maxlen" bytes.
|
|
|
1928 */
|
|
|
1929 void
|
|
|
1930 spell_cat_line(buf, line, maxlen)
|
|
|
1931 char_u *buf;
|
|
|
1932 char_u *line;
|
|
|
1933 int maxlen;
|
|
|
1934 {
|
|
|
1935 char_u *p;
|
|
|
1936 int n;
|
|
|
1937
|
|
|
1938 p = skipwhite(line);
|
|
|
1939 while (vim_strchr((char_u *)"*#/\"\t", *p) != NULL)
|
|
|
1940 p = skipwhite(p + 1);
|
|
|
1941
|
|
|
1942 if (*p != NUL)
|
|
|
1943 {
|
|
|
1944 *buf = ' ';
|
|
376
|
1945 vim_strncpy(buf + 1, line, maxlen - 2);
|
|
346
|
1946 n = p - line;
|
|
|
1947 if (n >= maxlen)
|
|
|
1948 n = maxlen - 1;
|
|
|
1949 vim_memset(buf + 1, ' ', n);
|
|
|
1950 }
|
|
236
|
1951 }
|
|
|
1952
|
|
|
1953 /*
|
|
307
|
1954 * Load word list(s) for "lang" from Vim spell file(s).
|
|
310
|
1955 * "lang" must be the language without the region: e.g., "en".
|
|
236
|
1956 */
|
|
307
|
1957 static void
|
|
236
|
1958 spell_load_lang(lang)
|
|
|
1959 char_u *lang;
|
|
|
1960 {
|
|
310
|
1961 char_u fname_enc[85];
|
|
236
|
1962 int r;
|
|
307
|
1963 char_u langcp[MAXWLEN + 1];
|
|
|
1964
|
|
310
|
1965 /* Copy the language name to pass it to spell_load_cb() as a cookie.
|
|
307
|
1966 * It's truncated when an error is detected. */
|
|
|
1967 STRCPY(langcp, lang);
|
|
|
1968
|
|
310
|
1969 /*
|
|
|
1970 * Find the first spell file for "lang" in 'runtimepath' and load it.
|
|
|
1971 */
|
|
|
1972 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5,
|
|
|
1973 "spell/%s.%s.spl", lang, spell_enc());
|
|
|
1974 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &langcp);
|
|
307
|
1975
|
|
|
1976 if (r == FAIL && *langcp != NUL)
|
|
|
1977 {
|
|
|
1978 /* Try loading the ASCII version. */
|
|
310
|
1979 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5,
|
|
272
|
1980 "spell/%s.ascii.spl", lang);
|
|
310
|
1981 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &langcp);
|
|
307
|
1982 }
|
|
|
1983
|
|
|
1984 if (r == FAIL)
|
|
483
|
1985 smsg((char_u *)_("Warning: Cannot find word list \"%s.%s.spl\" or \"%s.ascii.spl\""),
|
|
|
1986 lang, spell_enc(), lang);
|
|
310
|
1987 else if (*langcp != NUL)
|
|
|
1988 {
|
|
|
1989 /* Load all the additions. */
|
|
|
1990 STRCPY(fname_enc + STRLEN(fname_enc) - 3, "add.spl");
|
|
|
1991 do_in_runtimepath(fname_enc, TRUE, spell_load_cb, &langcp);
|
|
|
1992 }
|
|
|
1993 }
|
|
|
1994
|
|
|
1995 /*
|
|
|
1996 * Return the encoding used for spell checking: Use 'encoding', except that we
|
|
|
1997 * use "latin1" for "latin9". And limit to 60 characters (just in case).
|
|
|
1998 */
|
|
|
1999 static char_u *
|
|
|
2000 spell_enc()
|
|
|
2001 {
|
|
|
2002
|
|
|
2003 #ifdef FEAT_MBYTE
|
|
|
2004 if (STRLEN(p_enc) < 60 && STRCMP(p_enc, "iso-8859-15") != 0)
|
|
|
2005 return p_enc;
|
|
|
2006 #endif
|
|
|
2007 return (char_u *)"latin1";
|
|
236
|
2008 }
|
|
|
2009
|
|
|
2010 /*
|
|
385
|
2011 * Get the name of the .spl file for the internal wordlist into
|
|
|
2012 * "fname[MAXPATHL]".
|
|
|
2013 */
|
|
|
2014 static void
|
|
|
2015 int_wordlist_spl(fname)
|
|
|
2016 char_u *fname;
|
|
|
2017 {
|
|
|
2018 vim_snprintf((char *)fname, MAXPATHL, "%s.%s.spl",
|
|
|
2019 int_wordlist, spell_enc());
|
|
|
2020 }
|
|
|
2021
|
|
|
2022 /*
|
|
236
|
2023 * Allocate a new slang_T.
|
|
|
2024 * Caller must fill "sl_next".
|
|
|
2025 */
|
|
|
2026 static slang_T *
|
|
|
2027 slang_alloc(lang)
|
|
|
2028 char_u *lang;
|
|
|
2029 {
|
|
|
2030 slang_T *lp;
|
|
|
2031
|
|
300
|
2032 lp = (slang_T *)alloc_clear(sizeof(slang_T));
|
|
236
|
2033 if (lp != NULL)
|
|
|
2034 {
|
|
|
2035 lp->sl_name = vim_strsave(lang);
|
|
323
|
2036 ga_init2(&lp->sl_rep, sizeof(fromto_T), 10);
|
|
483
|
2037 lp->sl_compmax = MAXWLEN;
|
|
|
2038 lp->sl_compminlen = MAXWLEN;
|
|
|
2039 lp->sl_compsylmax = MAXWLEN;
|
|
236
|
2040 }
|
|
|
2041 return lp;
|
|
|
2042 }
|
|
|
2043
|
|
|
2044 /*
|
|
|
2045 * Free the contents of an slang_T and the structure itself.
|
|
|
2046 */
|
|
|
2047 static void
|
|
|
2048 slang_free(lp)
|
|
|
2049 slang_T *lp;
|
|
|
2050 {
|
|
|
2051 vim_free(lp->sl_name);
|
|
310
|
2052 vim_free(lp->sl_fname);
|
|
|
2053 slang_clear(lp);
|
|
|
2054 vim_free(lp);
|
|
|
2055 }
|
|
|
2056
|
|
|
2057 /*
|
|
|
2058 * Clear an slang_T so that the file can be reloaded.
|
|
|
2059 */
|
|
|
2060 static void
|
|
|
2061 slang_clear(lp)
|
|
|
2062 slang_T *lp;
|
|
|
2063 {
|
|
339
|
2064 garray_T *gap;
|
|
|
2065 fromto_T *ftp;
|
|
344
|
2066 salitem_T *smp;
|
|
339
|
2067 int i;
|
|
323
|
2068
|
|
300
|
2069 vim_free(lp->sl_fbyts);
|
|
310
|
2070 lp->sl_fbyts = NULL;
|
|
300
|
2071 vim_free(lp->sl_kbyts);
|
|
310
|
2072 lp->sl_kbyts = NULL;
|
|
339
|
2073 vim_free(lp->sl_pbyts);
|
|
|
2074 lp->sl_pbyts = NULL;
|
|
|
2075
|
|
300
|
2076 vim_free(lp->sl_fidxs);
|
|
310
|
2077 lp->sl_fidxs = NULL;
|
|
300
|
2078 vim_free(lp->sl_kidxs);
|
|
310
|
2079 lp->sl_kidxs = NULL;
|
|
339
|
2080 vim_free(lp->sl_pidxs);
|
|
|
2081 lp->sl_pidxs = NULL;
|
|
323
|
2082
|
|
344
|
2083 gap = &lp->sl_rep;
|
|
|
2084 while (gap->ga_len > 0)
|
|
323
|
2085 {
|
|
344
|
2086 ftp = &((fromto_T *)gap->ga_data)[--gap->ga_len];
|
|
|
2087 vim_free(ftp->ft_from);
|
|
|
2088 vim_free(ftp->ft_to);
|
|
323
|
2089 }
|
|
344
|
2090 ga_clear(gap);
|
|
|
2091
|
|
|
2092 gap = &lp->sl_sal;
|
|
375
|
2093 if (lp->sl_sofo)
|
|
376
|
2094 {
|
|
|
2095 /* "ga_len" is set to 1 without adding an item for latin1 */
|
|
|
2096 if (gap->ga_data != NULL)
|
|
|
2097 /* SOFOFROM and SOFOTO items: free lists of wide characters. */
|
|
|
2098 for (i = 0; i < gap->ga_len; ++i)
|
|
|
2099 vim_free(((int **)gap->ga_data)[i]);
|
|
|
2100 }
|
|
375
|
2101 else
|
|
|
2102 /* SAL items: free salitem_T items */
|
|
|
2103 while (gap->ga_len > 0)
|
|
|
2104 {
|
|
|
2105 smp = &((salitem_T *)gap->ga_data)[--gap->ga_len];
|
|
|
2106 vim_free(smp->sm_lead);
|
|
|
2107 /* Don't free sm_oneof and sm_rules, they point into sm_lead. */
|
|
|
2108 vim_free(smp->sm_to);
|
|
|
2109 #ifdef FEAT_MBYTE
|
|
|
2110 vim_free(smp->sm_lead_w);
|
|
|
2111 vim_free(smp->sm_oneof_w);
|
|
|
2112 vim_free(smp->sm_to_w);
|
|
|
2113 #endif
|
|
|
2114 }
|
|
344
|
2115 ga_clear(gap);
|
|
323
|
2116
|
|
339
|
2117 for (i = 0; i < lp->sl_prefixcnt; ++i)
|
|
|
2118 vim_free(lp->sl_prefprog[i]);
|
|
376
|
2119 lp->sl_prefixcnt = 0;
|
|
339
|
2120 vim_free(lp->sl_prefprog);
|
|
376
|
2121 lp->sl_prefprog = NULL;
|
|
|
2122
|
|
|
2123 vim_free(lp->sl_midword);
|
|
|
2124 lp->sl_midword = NULL;
|
|
339
|
2125
|
|
483
|
2126 vim_free(lp->sl_compprog);
|
|
|
2127 vim_free(lp->sl_compstartflags);
|
|
485
|
2128 vim_free(lp->sl_compallflags);
|
|
483
|
2129 lp->sl_compprog = NULL;
|
|
|
2130 lp->sl_compstartflags = NULL;
|
|
485
|
2131 lp->sl_compallflags = NULL;
|
|
483
|
2132
|
|
|
2133 vim_free(lp->sl_syllable);
|
|
|
2134 lp->sl_syllable = NULL;
|
|
|
2135 ga_clear(&lp->sl_syl_items);
|
|
481
|
2136
|
|
330
|
2137 #ifdef FEAT_MBYTE
|
|
|
2138 {
|
|
|
2139 int todo = lp->sl_map_hash.ht_used;
|
|
|
2140 hashitem_T *hi;
|
|
|
2141
|
|
|
2142 for (hi = lp->sl_map_hash.ht_array; todo > 0; ++hi)
|
|
|
2143 if (!HASHITEM_EMPTY(hi))
|
|
|
2144 {
|
|
|
2145 --todo;
|
|
|
2146 vim_free(hi->hi_key);
|
|
|
2147 }
|
|
|
2148 }
|
|
|
2149 hash_clear(&lp->sl_map_hash);
|
|
|
2150 #endif
|
|
483
|
2151
|
|
|
2152 lp->sl_compmax = MAXWLEN;
|
|
|
2153 lp->sl_compminlen = MAXWLEN;
|
|
|
2154 lp->sl_compsylmax = MAXWLEN;
|
|
|
2155 lp->sl_regions[0] = NUL;
|
|
236
|
2156 }
|
|
|
2157
|
|
|
2158 /*
|
|
307
|
2159 * Load one spell file and store the info into a slang_T.
|
|
236
|
2160 * Invoked through do_in_runtimepath().
|
|
|
2161 */
|
|
|
2162 static void
|
|
310
|
2163 spell_load_cb(fname, cookie)
|
|
236
|
2164 char_u *fname;
|
|
307
|
2165 void *cookie; /* points to the language name */
|
|
236
|
2166 {
|
|
323
|
2167 (void)spell_load_file(fname, (char_u *)cookie, NULL, FALSE);
|
|
310
|
2168 }
|
|
|
2169
|
|
|
2170 /*
|
|
|
2171 * Load one spell file and store the info into a slang_T.
|
|
|
2172 *
|
|
|
2173 * This is invoked in two ways:
|
|
|
2174 * - From spell_load_cb() to load a spell file for the first time. "lang" is
|
|
|
2175 * the language name, "old_lp" is NULL. Will allocate an slang_T.
|
|
|
2176 * - To reload a spell file that was changed. "lang" is NULL and "old_lp"
|
|
|
2177 * points to the existing slang_T.
|
|
323
|
2178 * Returns the slang_T the spell file was loaded into. NULL for error.
|
|
310
|
2179 */
|
|
323
|
2180 static slang_T *
|
|
|
2181 spell_load_file(fname, lang, old_lp, silent)
|
|
310
|
2182 char_u *fname;
|
|
|
2183 char_u *lang;
|
|
|
2184 slang_T *old_lp;
|
|
323
|
2185 int silent; /* no error if file doesn't exist */
|
|
310
|
2186 {
|
|
236
|
2187 FILE *fd;
|
|
483
|
2188 char_u buf[VIMSPELLMAGICL];
|
|
236
|
2189 char_u *p;
|
|
339
|
2190 char_u *bp;
|
|
|
2191 idx_T *ip;
|
|
236
|
2192 int i;
|
|
339
|
2193 int n;
|
|
300
|
2194 int len;
|
|
236
|
2195 int round;
|
|
|
2196 char_u *save_sourcing_name = sourcing_name;
|
|
|
2197 linenr_T save_sourcing_lnum = sourcing_lnum;
|
|
307
|
2198 slang_T *lp = NULL;
|
|
324
|
2199 idx_T idx;
|
|
344
|
2200 int c = 0;
|
|
483
|
2201 int res;
|
|
236
|
2202
|
|
310
|
2203 fd = mch_fopen((char *)fname, "r");
|
|
236
|
2204 if (fd == NULL)
|
|
|
2205 {
|
|
323
|
2206 if (!silent)
|
|
|
2207 EMSG2(_(e_notopen), fname);
|
|
|
2208 else if (p_verbose > 2)
|
|
|
2209 {
|
|
|
2210 verbose_enter();
|
|
|
2211 smsg((char_u *)e_notopen, fname);
|
|
|
2212 verbose_leave();
|
|
|
2213 }
|
|
255
|
2214 goto endFAIL;
|
|
236
|
2215 }
|
|
310
|
2216 if (p_verbose > 2)
|
|
|
2217 {
|
|
|
2218 verbose_enter();
|
|
|
2219 smsg((char_u *)_("Reading spell file \"%s\""), fname);
|
|
|
2220 verbose_leave();
|
|
|
2221 }
|
|
|
2222
|
|
|
2223 if (old_lp == NULL)
|
|
|
2224 {
|
|
|
2225 lp = slang_alloc(lang);
|
|
|
2226 if (lp == NULL)
|
|
|
2227 goto endFAIL;
|
|
|
2228
|
|
|
2229 /* Remember the file name, used to reload the file when it's updated. */
|
|
|
2230 lp->sl_fname = vim_strsave(fname);
|
|
|
2231 if (lp->sl_fname == NULL)
|
|
|
2232 goto endFAIL;
|
|
|
2233
|
|
|
2234 /* Check for .add.spl. */
|
|
|
2235 lp->sl_add = strstr((char *)gettail(fname), ".add.") != NULL;
|
|
|
2236 }
|
|
|
2237 else
|
|
|
2238 lp = old_lp;
|
|
307
|
2239
|
|
236
|
2240 /* Set sourcing_name, so that error messages mention the file name. */
|
|
|
2241 sourcing_name = fname;
|
|
|
2242 sourcing_lnum = 0;
|
|
|
2243
|
|
339
|
2244 /* <HEADER>: <fileID>
|
|
|
2245 */
|
|
236
|
2246 for (i = 0; i < VIMSPELLMAGICL; ++i)
|
|
|
2247 buf[i] = getc(fd); /* <fileID> */
|
|
|
2248 if (STRNCMP(buf, VIMSPELLMAGIC, VIMSPELLMAGICL) != 0)
|
|
|
2249 {
|
|
483
|
2250 EMSG(_("E757: This does not look like a spell file"));
|
|
381
|
2251 goto endFAIL;
|
|
483
|
2252 }
|
|
|
2253 c = getc(fd); /* <versionnr> */
|
|
|
2254 if (c < VIMSPELLVERSION)
|
|
|
2255 {
|
|
|
2256 EMSG(_("E771: Old spell file, needs to be updated"));
|
|
381
|
2257 goto endFAIL;
|
|
|
2258 }
|
|
483
|
2259 else if (c > VIMSPELLVERSION)
|
|
|
2260 {
|
|
|
2261 EMSG(_("E772: Spell file is for newer version of Vim"));
|
|
481
|
2262 goto endFAIL;
|
|
483
|
2263 }
|
|
|
2264
|
|
|
2265
|
|
|
2266 /*
|
|
|
2267 * <SECTIONS>: <section> ... <sectionend>
|
|
|
2268 * <section>: <sectionID> <sectionflags> <sectionlen> (section contents)
|
|
|
2269 */
|
|
|
2270 for (;;)
|
|
|
2271 {
|
|
|
2272 n = getc(fd); /* <sectionID> or <sectionend> */
|
|
|
2273 if (n == SN_END)
|
|
|
2274 break;
|
|
|
2275 c = getc(fd); /* <sectionflags> */
|
|
|
2276 len = (getc(fd) << 24) + (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
2277 /* <sectionlen> */
|
|
|
2278 if (len < 0)
|
|
|
2279 goto truncerr;
|
|
|
2280
|
|
|
2281 res = 0;
|
|
|
2282 switch (n)
|
|
|
2283 {
|
|
|
2284 case SN_REGION:
|
|
|
2285 res = read_region_section(fd, lp, len);
|
|
|
2286 break;
|
|
|
2287
|
|
|
2288 case SN_CHARFLAGS:
|
|
|
2289 res = read_charflags_section(fd);
|
|
|
2290 break;
|
|
|
2291
|
|
|
2292 case SN_MIDWORD:
|
|
|
2293 lp->sl_midword = read_string(fd, len); /* <midword> */
|
|
|
2294 if (lp->sl_midword == NULL)
|
|
|
2295 goto endFAIL;
|
|
|
2296 break;
|
|
|
2297
|
|
|
2298 case SN_PREFCOND:
|
|
|
2299 res = read_prefcond_section(fd, lp);
|
|
|
2300 break;
|
|
|
2301
|
|
|
2302 case SN_REP:
|
|
|
2303 res = read_rep_section(fd, lp);
|
|
|
2304 break;
|
|
|
2305
|
|
|
2306 case SN_SAL:
|
|
|
2307 res = read_sal_section(fd, lp);
|
|
|
2308 break;
|
|
|
2309
|
|
|
2310 case SN_SOFO:
|
|
|
2311 res = read_sofo_section(fd, lp);
|
|
|
2312 break;
|
|
|
2313
|
|
|
2314 case SN_MAP:
|
|
|
2315 p = read_string(fd, len); /* <mapstr> */
|
|
|
2316 if (p == NULL)
|
|
|
2317 goto endFAIL;
|
|
|
2318 set_map_str(lp, p);
|
|
|
2319 vim_free(p);
|
|
|
2320 break;
|
|
|
2321
|
|
|
2322 case SN_COMPOUND:
|
|
|
2323 res = read_compound(fd, lp, len);
|
|
|
2324 break;
|
|
|
2325
|
|
492
|
2326 case SN_NOBREAK:
|
|
|
2327 lp->sl_nobreak = TRUE;
|
|
|
2328 break;
|
|
|
2329
|
|
483
|
2330 case SN_SYLLABLE:
|
|
|
2331 lp->sl_syllable = read_string(fd, len); /* <syllable> */
|
|
|
2332 if (lp->sl_syllable == NULL)
|
|
|
2333 goto endFAIL;
|
|
|
2334 if (init_syl_tab(lp) == FAIL)
|
|
|
2335 goto endFAIL;
|
|
|
2336 break;
|
|
|
2337
|
|
|
2338 default:
|
|
|
2339 /* Unsupported section. When it's required give an error
|
|
|
2340 * message. When it's not required skip the contents. */
|
|
|
2341 if (c & SNF_REQUIRED)
|
|
|
2342 {
|
|
|
2343 EMSG(_("E770: Unsupported section in spell file"));
|
|
|
2344 goto endFAIL;
|
|
|
2345 }
|
|
|
2346 while (--len >= 0)
|
|
|
2347 if (getc(fd) < 0)
|
|
|
2348 goto truncerr;
|
|
|
2349 break;
|
|
|
2350 }
|
|
|
2351 if (res == SP_FORMERROR)
|
|
|
2352 {
|
|
|
2353 formerr:
|
|
|
2354 EMSG(_(e_format));
|
|
381
|
2355 goto endFAIL;
|
|
344
|
2356 }
|
|
483
|
2357 if (res == SP_TRUNCERROR)
|
|
|
2358 {
|
|
|
2359 truncerr:
|
|
|
2360 EMSG(_(e_spell_trunc));
|
|
375
|
2361 goto endFAIL;
|
|
483
|
2362 }
|
|
|
2363 if (res == SP_ERROR)
|
|
|
2364 goto endFAIL;
|
|
389
|
2365 }
|
|
236
|
2366
|
|
300
|
2367 /* round 1: <LWORDTREE>
|
|
339
|
2368 * round 2: <KWORDTREE>
|
|
|
2369 * round 3: <PREFIXTREE> */
|
|
|
2370 for (round = 1; round <= 3; ++round)
|
|
236
|
2371 {
|
|
300
|
2372 /* The tree size was computed when writing the file, so that we can
|
|
|
2373 * allocate it as one long block. <nodecount> */
|
|
|
2374 len = (getc(fd) << 24) + (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
2375 if (len < 0)
|
|
|
2376 goto truncerr;
|
|
|
2377 if (len > 0)
|
|
236
|
2378 {
|
|
300
|
2379 /* Allocate the byte array. */
|
|
339
|
2380 bp = lalloc((long_u)len, TRUE);
|
|
|
2381 if (bp == NULL)
|
|
300
|
2382 goto endFAIL;
|
|
|
2383 if (round == 1)
|
|
339
|
2384 lp->sl_fbyts = bp;
|
|
|
2385 else if (round == 2)
|
|
|
2386 lp->sl_kbyts = bp;
|
|
300
|
2387 else
|
|
339
|
2388 lp->sl_pbyts = bp;
|
|
236
|
2389
|
|
300
|
2390 /* Allocate the index array. */
|
|
339
|
2391 ip = (idx_T *)lalloc_clear((long_u)(len * sizeof(int)), TRUE);
|
|
|
2392 if (ip == NULL)
|
|
300
|
2393 goto endFAIL;
|
|
|
2394 if (round == 1)
|
|
339
|
2395 lp->sl_fidxs = ip;
|
|
|
2396 else if (round == 2)
|
|
|
2397 lp->sl_kidxs = ip;
|
|
300
|
2398 else
|
|
339
|
2399 lp->sl_pidxs = ip;
|
|
300
|
2400
|
|
|
2401 /* Read the tree and store it in the array. */
|
|
339
|
2402 idx = read_tree(fd, bp, ip, len, 0, round == 3, lp->sl_prefixcnt);
|
|
324
|
2403 if (idx == -1)
|
|
300
|
2404 goto truncerr;
|
|
324
|
2405 if (idx < 0)
|
|
236
|
2406 goto formerr;
|
|
|
2407 }
|
|
300
|
2408 }
|
|
243
|
2409
|
|
310
|
2410 /* For a new file link it in the list of spell files. */
|
|
|
2411 if (old_lp == NULL)
|
|
|
2412 {
|
|
|
2413 lp->sl_next = first_lang;
|
|
|
2414 first_lang = lp;
|
|
|
2415 }
|
|
307
|
2416
|
|
255
|
2417 goto endOK;
|
|
|
2418
|
|
|
2419 endFAIL:
|
|
310
|
2420 if (lang != NULL)
|
|
|
2421 /* truncating the name signals the error to spell_load_lang() */
|
|
|
2422 *lang = NUL;
|
|
|
2423 if (lp != NULL && old_lp == NULL)
|
|
323
|
2424 {
|
|
307
|
2425 slang_free(lp);
|
|
323
|
2426 lp = NULL;
|
|
|
2427 }
|
|
255
|
2428
|
|
|
2429 endOK:
|
|
236
|
2430 if (fd != NULL)
|
|
|
2431 fclose(fd);
|
|
|
2432 sourcing_name = save_sourcing_name;
|
|
|
2433 sourcing_lnum = save_sourcing_lnum;
|
|
323
|
2434
|
|
|
2435 return lp;
|
|
236
|
2436 }
|
|
|
2437
|
|
376
|
2438 /*
|
|
|
2439 * Read a length field from "fd" in "cnt_bytes" bytes.
|
|
381
|
2440 * Allocate memory, read the string into it and add a NUL at the end.
|
|
376
|
2441 * Returns NULL when the count is zero.
|
|
483
|
2442 * Sets "*cntp" to SP_*ERROR when there is an error, length of the result
|
|
|
2443 * otherwise.
|
|
376
|
2444 */
|
|
|
2445 static char_u *
|
|
389
|
2446 read_cnt_string(fd, cnt_bytes, cntp)
|
|
376
|
2447 FILE *fd;
|
|
|
2448 int cnt_bytes;
|
|
389
|
2449 int *cntp;
|
|
376
|
2450 {
|
|
|
2451 int cnt = 0;
|
|
|
2452 int i;
|
|
|
2453 char_u *str;
|
|
|
2454
|
|
|
2455 /* read the length bytes, MSB first */
|
|
|
2456 for (i = 0; i < cnt_bytes; ++i)
|
|
|
2457 cnt = (cnt << 8) + getc(fd);
|
|
|
2458 if (cnt < 0)
|
|
|
2459 {
|
|
483
|
2460 *cntp = SP_TRUNCERROR;
|
|
376
|
2461 return NULL;
|
|
|
2462 }
|
|
389
|
2463 *cntp = cnt;
|
|
|
2464 if (cnt == 0)
|
|
|
2465 return NULL; /* nothing to read, return NULL */
|
|
376
|
2466
|
|
483
|
2467 str = read_string(fd, cnt);
|
|
|
2468 if (str == NULL)
|
|
|
2469 *cntp = SP_ERROR;
|
|
|
2470 return str;
|
|
|
2471 }
|
|
|
2472
|
|
|
2473 /*
|
|
|
2474 * Read a string of length "cnt" from "fd" into allocated memory.
|
|
|
2475 * Returns NULL when out of memory.
|
|
|
2476 */
|
|
|
2477 static char_u *
|
|
|
2478 read_string(fd, cnt)
|
|
|
2479 FILE *fd;
|
|
|
2480 int cnt;
|
|
|
2481 {
|
|
|
2482 char_u *str;
|
|
|
2483 int i;
|
|
|
2484
|
|
376
|
2485 /* allocate memory */
|
|
|
2486 str = alloc((unsigned)cnt + 1);
|
|
483
|
2487 if (str != NULL)
|
|
|
2488 {
|
|
|
2489 /* Read the string. Doesn't check for truncated file. */
|
|
|
2490 for (i = 0; i < cnt; ++i)
|
|
|
2491 str[i] = getc(fd);
|
|
|
2492 str[i] = NUL;
|
|
|
2493 }
|
|
|
2494 return str;
|
|
|
2495 }
|
|
|
2496
|
|
|
2497 /*
|
|
|
2498 * Read SN_REGION: <regionname> ...
|
|
|
2499 * Return SP_*ERROR flags.
|
|
|
2500 */
|
|
|
2501 static int
|
|
|
2502 read_region_section(fd, lp, len)
|
|
|
2503 FILE *fd;
|
|
|
2504 slang_T *lp;
|
|
|
2505 int len;
|
|
|
2506 {
|
|
|
2507 int i;
|
|
|
2508
|
|
|
2509 if (len > 16)
|
|
|
2510 return SP_FORMERROR;
|
|
|
2511 for (i = 0; i < len; ++i)
|
|
|
2512 lp->sl_regions[i] = getc(fd); /* <regionname> */
|
|
|
2513 lp->sl_regions[len] = NUL;
|
|
|
2514 return 0;
|
|
|
2515 }
|
|
|
2516
|
|
|
2517 /*
|
|
|
2518 * Read SN_CHARFLAGS section: <charflagslen> <charflags>
|
|
|
2519 * <folcharslen> <folchars>
|
|
|
2520 * Return SP_*ERROR flags.
|
|
|
2521 */
|
|
|
2522 static int
|
|
|
2523 read_charflags_section(fd)
|
|
|
2524 FILE *fd;
|
|
|
2525 {
|
|
|
2526 char_u *flags;
|
|
|
2527 char_u *fol;
|
|
|
2528 int flagslen, follen;
|
|
|
2529
|
|
|
2530 /* <charflagslen> <charflags> */
|
|
|
2531 flags = read_cnt_string(fd, 1, &flagslen);
|
|
|
2532 if (flagslen < 0)
|
|
|
2533 return flagslen;
|
|
|
2534
|
|
|
2535 /* <folcharslen> <folchars> */
|
|
|
2536 fol = read_cnt_string(fd, 2, &follen);
|
|
|
2537 if (follen < 0)
|
|
|
2538 {
|
|
|
2539 vim_free(flags);
|
|
|
2540 return follen;
|
|
|
2541 }
|
|
|
2542
|
|
|
2543 /* Set the word-char flags and fill SPELL_ISUPPER() table. */
|
|
|
2544 if (flags != NULL && fol != NULL)
|
|
|
2545 set_spell_charflags(flags, flagslen, fol);
|
|
|
2546
|
|
|
2547 vim_free(flags);
|
|
|
2548 vim_free(fol);
|
|
|
2549
|
|
|
2550 /* When <charflagslen> is zero then <fcharlen> must also be zero. */
|
|
|
2551 if ((flags == NULL) != (fol == NULL))
|
|
|
2552 return SP_FORMERROR;
|
|
|
2553 return 0;
|
|
|
2554 }
|
|
|
2555
|
|
|
2556 /*
|
|
|
2557 * Read SN_PREFCOND section.
|
|
|
2558 * Return SP_*ERROR flags.
|
|
|
2559 */
|
|
|
2560 static int
|
|
|
2561 read_prefcond_section(fd, lp)
|
|
|
2562 FILE *fd;
|
|
|
2563 slang_T *lp;
|
|
|
2564 {
|
|
|
2565 int cnt;
|
|
|
2566 int i;
|
|
|
2567 int n;
|
|
|
2568 char_u *p;
|
|
|
2569 char_u buf[MAXWLEN + 1];
|
|
|
2570
|
|
|
2571 /* <prefcondcnt> <prefcond> ... */
|
|
|
2572 cnt = (getc(fd) << 8) + getc(fd); /* <prefcondcnt> */
|
|
|
2573 if (cnt <= 0)
|
|
|
2574 return SP_FORMERROR;
|
|
|
2575
|
|
|
2576 lp->sl_prefprog = (regprog_T **)alloc_clear(
|
|
|
2577 (unsigned)sizeof(regprog_T *) * cnt);
|
|
|
2578 if (lp->sl_prefprog == NULL)
|
|
|
2579 return SP_ERROR;
|
|
|
2580 lp->sl_prefixcnt = cnt;
|
|
|
2581
|
|
376
|
2582 for (i = 0; i < cnt; ++i)
|
|
483
|
2583 {
|
|
|
2584 /* <prefcond> : <condlen> <condstr> */
|
|
|
2585 n = getc(fd); /* <condlen> */
|
|
|
2586 if (n < 0 || n >= MAXWLEN)
|
|
|
2587 return SP_FORMERROR;
|
|
|
2588
|
|
|
2589 /* When <condlen> is zero we have an empty condition. Otherwise
|
|
|
2590 * compile the regexp program used to check for the condition. */
|
|
|
2591 if (n > 0)
|
|
|
2592 {
|
|
|
2593 buf[0] = '^'; /* always match at one position only */
|
|
|
2594 p = buf + 1;
|
|
|
2595 while (n-- > 0)
|
|
|
2596 *p++ = getc(fd); /* <condstr> */
|
|
|
2597 *p = NUL;
|
|
|
2598 lp->sl_prefprog[i] = vim_regcomp(buf, RE_MAGIC + RE_STRING);
|
|
|
2599 }
|
|
|
2600 }
|
|
|
2601 return 0;
|
|
|
2602 }
|
|
|
2603
|
|
|
2604 /*
|
|
|
2605 * Read REP items section from "fd": <repcount> <rep> ...
|
|
|
2606 * Return SP_*ERROR flags.
|
|
|
2607 */
|
|
|
2608 static int
|
|
|
2609 read_rep_section(fd, slang)
|
|
|
2610 FILE *fd;
|
|
|
2611 slang_T *slang;
|
|
|
2612 {
|
|
|
2613 int cnt;
|
|
|
2614 garray_T *gap;
|
|
|
2615 fromto_T *ftp;
|
|
|
2616 short *first;
|
|
|
2617 int i;
|
|
|
2618
|
|
|
2619 cnt = (getc(fd) << 8) + getc(fd); /* <repcount> */
|
|
|
2620 if (cnt < 0)
|
|
|
2621 return SP_TRUNCERROR;
|
|
|
2622
|
|
|
2623 gap = &slang->sl_rep;
|
|
|
2624 if (ga_grow(gap, cnt) == FAIL)
|
|
|
2625 return SP_ERROR;
|
|
|
2626
|
|
|
2627 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */
|
|
|
2628 for (; gap->ga_len < cnt; ++gap->ga_len)
|
|
|
2629 {
|
|
|
2630 ftp = &((fromto_T *)gap->ga_data)[gap->ga_len];
|
|
|
2631 ftp->ft_from = read_cnt_string(fd, 1, &i);
|
|
|
2632 if (i < 0)
|
|
|
2633 return i;
|
|
|
2634 if (i == 0)
|
|
|
2635 return SP_FORMERROR;
|
|
|
2636 ftp->ft_to = read_cnt_string(fd, 1, &i);
|
|
|
2637 if (i <= 0)
|
|
|
2638 {
|
|
|
2639 vim_free(ftp->ft_from);
|
|
|
2640 if (i < 0)
|
|
|
2641 return i;
|
|
|
2642 return SP_FORMERROR;
|
|
|
2643 }
|
|
|
2644 }
|
|
|
2645
|
|
|
2646 /* Fill the first-index table. */
|
|
|
2647 first = slang->sl_rep_first;
|
|
|
2648 for (i = 0; i < 256; ++i)
|
|
|
2649 first[i] = -1;
|
|
|
2650 for (i = 0; i < gap->ga_len; ++i)
|
|
|
2651 {
|
|
|
2652 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
2653 if (first[*ftp->ft_from] == -1)
|
|
|
2654 first[*ftp->ft_from] = i;
|
|
|
2655 }
|
|
|
2656 return 0;
|
|
|
2657 }
|
|
|
2658
|
|
|
2659 /*
|
|
|
2660 * Read SN_SAL section: <salflags> <salcount> <sal> ...
|
|
|
2661 * Return SP_*ERROR flags.
|
|
|
2662 */
|
|
|
2663 static int
|
|
|
2664 read_sal_section(fd, slang)
|
|
|
2665 FILE *fd;
|
|
|
2666 slang_T *slang;
|
|
|
2667 {
|
|
|
2668 int i;
|
|
|
2669 int cnt;
|
|
|
2670 garray_T *gap;
|
|
|
2671 salitem_T *smp;
|
|
|
2672 int ccnt;
|
|
|
2673 char_u *p;
|
|
485
|
2674 int c = NUL;
|
|
483
|
2675
|
|
|
2676 slang->sl_sofo = FALSE;
|
|
|
2677
|
|
|
2678 i = getc(fd); /* <salflags> */
|
|
|
2679 if (i & SAL_F0LLOWUP)
|
|
|
2680 slang->sl_followup = TRUE;
|
|
|
2681 if (i & SAL_COLLAPSE)
|
|
|
2682 slang->sl_collapse = TRUE;
|
|
|
2683 if (i & SAL_REM_ACCENTS)
|
|
|
2684 slang->sl_rem_accents = TRUE;
|
|
|
2685
|
|
|
2686 cnt = (getc(fd) << 8) + getc(fd); /* <salcount> */
|
|
|
2687 if (cnt < 0)
|
|
|
2688 return SP_TRUNCERROR;
|
|
|
2689
|
|
|
2690 gap = &slang->sl_sal;
|
|
|
2691 ga_init2(gap, sizeof(salitem_T), 10);
|
|
|
2692 if (ga_grow(gap, cnt) == FAIL)
|
|
|
2693 return SP_ERROR;
|
|
|
2694
|
|
|
2695 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */
|
|
|
2696 for (; gap->ga_len < cnt; ++gap->ga_len)
|
|
|
2697 {
|
|
|
2698 smp = &((salitem_T *)gap->ga_data)[gap->ga_len];
|
|
|
2699 ccnt = getc(fd); /* <salfromlen> */
|
|
|
2700 if (ccnt < 0)
|
|
|
2701 return SP_TRUNCERROR;
|
|
|
2702 if ((p = alloc(ccnt + 2)) == NULL)
|
|
|
2703 return SP_ERROR;
|
|
|
2704 smp->sm_lead = p;
|
|
|
2705
|
|
|
2706 /* Read up to the first special char into sm_lead. */
|
|
|
2707 for (i = 0; i < ccnt; ++i)
|
|
|
2708 {
|
|
|
2709 c = getc(fd); /* <salfrom> */
|
|
|
2710 if (vim_strchr((char_u *)"0123456789(-<^$", c) != NULL)
|
|
|
2711 break;
|
|
|
2712 *p++ = c;
|
|
|
2713 }
|
|
|
2714 smp->sm_leadlen = p - smp->sm_lead;
|
|
|
2715 *p++ = NUL;
|
|
|
2716
|
|
|
2717 /* Put (abc) chars in sm_oneof, if any. */
|
|
|
2718 if (c == '(')
|
|
|
2719 {
|
|
|
2720 smp->sm_oneof = p;
|
|
|
2721 for (++i; i < ccnt; ++i)
|
|
|
2722 {
|
|
|
2723 c = getc(fd); /* <salfrom> */
|
|
|
2724 if (c == ')')
|
|
|
2725 break;
|
|
|
2726 *p++ = c;
|
|
|
2727 }
|
|
|
2728 *p++ = NUL;
|
|
|
2729 if (++i < ccnt)
|
|
|
2730 c = getc(fd);
|
|
|
2731 }
|
|
|
2732 else
|
|
|
2733 smp->sm_oneof = NULL;
|
|
|
2734
|
|
|
2735 /* Any following chars go in sm_rules. */
|
|
|
2736 smp->sm_rules = p;
|
|
|
2737 if (i < ccnt)
|
|
|
2738 /* store the char we got while checking for end of sm_lead */
|
|
|
2739 *p++ = c;
|
|
|
2740 for (++i; i < ccnt; ++i)
|
|
|
2741 *p++ = getc(fd); /* <salfrom> */
|
|
|
2742 *p++ = NUL;
|
|
|
2743
|
|
|
2744 /* <saltolen> <salto> */
|
|
|
2745 smp->sm_to = read_cnt_string(fd, 1, &ccnt);
|
|
|
2746 if (ccnt < 0)
|
|
|
2747 {
|
|
|
2748 vim_free(smp->sm_lead);
|
|
|
2749 return ccnt;
|
|
|
2750 }
|
|
|
2751
|
|
|
2752 #ifdef FEAT_MBYTE
|
|
|
2753 if (has_mbyte)
|
|
|
2754 {
|
|
|
2755 /* convert the multi-byte strings to wide char strings */
|
|
|
2756 smp->sm_lead_w = mb_str2wide(smp->sm_lead);
|
|
|
2757 smp->sm_leadlen = mb_charlen(smp->sm_lead);
|
|
|
2758 if (smp->sm_oneof == NULL)
|
|
|
2759 smp->sm_oneof_w = NULL;
|
|
|
2760 else
|
|
|
2761 smp->sm_oneof_w = mb_str2wide(smp->sm_oneof);
|
|
|
2762 if (smp->sm_to == NULL)
|
|
|
2763 smp->sm_to_w = NULL;
|
|
|
2764 else
|
|
|
2765 smp->sm_to_w = mb_str2wide(smp->sm_to);
|
|
|
2766 if (smp->sm_lead_w == NULL
|
|
|
2767 || (smp->sm_oneof_w == NULL && smp->sm_oneof != NULL)
|
|
|
2768 || (smp->sm_to_w == NULL && smp->sm_to != NULL))
|
|
|
2769 {
|
|
|
2770 vim_free(smp->sm_lead);
|
|
|
2771 vim_free(smp->sm_to);
|
|
|
2772 vim_free(smp->sm_lead_w);
|
|
|
2773 vim_free(smp->sm_oneof_w);
|
|
|
2774 vim_free(smp->sm_to_w);
|
|
|
2775 return SP_ERROR;
|
|
|
2776 }
|
|
|
2777 }
|
|
|
2778 #endif
|
|
|
2779 }
|
|
|
2780
|
|
|
2781 /* Fill the first-index table. */
|
|
|
2782 set_sal_first(slang);
|
|
|
2783
|
|
|
2784 return 0;
|
|
|
2785 }
|
|
|
2786
|
|
|
2787 /*
|
|
|
2788 * SN_SOFO: <sofofromlen> <sofofrom> <sofotolen> <sofoto>
|
|
|
2789 * Return SP_*ERROR flags.
|
|
|
2790 */
|
|
|
2791 static int
|
|
|
2792 read_sofo_section(fd, slang)
|
|
|
2793 FILE *fd;
|
|
|
2794 slang_T *slang;
|
|
|
2795 {
|
|
|
2796 int cnt;
|
|
|
2797 char_u *from, *to;
|
|
|
2798 int res;
|
|
|
2799
|
|
|
2800 slang->sl_sofo = TRUE;
|
|
|
2801
|
|
|
2802 /* <sofofromlen> <sofofrom> */
|
|
|
2803 from = read_cnt_string(fd, 2, &cnt);
|
|
|
2804 if (cnt < 0)
|
|
|
2805 return cnt;
|
|
|
2806
|
|
|
2807 /* <sofotolen> <sofoto> */
|
|
|
2808 to = read_cnt_string(fd, 2, &cnt);
|
|
|
2809 if (cnt < 0)
|
|
|
2810 {
|
|
|
2811 vim_free(from);
|
|
|
2812 return cnt;
|
|
|
2813 }
|
|
|
2814
|
|
|
2815 /* Store the info in slang->sl_sal and/or slang->sl_sal_first. */
|
|
|
2816 if (from != NULL && to != NULL)
|
|
|
2817 res = set_sofo(slang, from, to);
|
|
|
2818 else if (from != NULL || to != NULL)
|
|
|
2819 res = SP_FORMERROR; /* only one of two strings is an error */
|
|
|
2820 else
|
|
|
2821 res = 0;
|
|
|
2822
|
|
|
2823 vim_free(from);
|
|
|
2824 vim_free(to);
|
|
|
2825 return res;
|
|
|
2826 }
|
|
|
2827
|
|
|
2828 /*
|
|
|
2829 * Read the compound section from the .spl file:
|
|
|
2830 * <compmax> <compminlen> <compsylmax> <compflags>
|
|
|
2831 * Returns SP_*ERROR flags.
|
|
|
2832 */
|
|
|
2833 static int
|
|
|
2834 read_compound(fd, slang, len)
|
|
|
2835 FILE *fd;
|
|
|
2836 slang_T *slang;
|
|
|
2837 int len;
|
|
|
2838 {
|
|
|
2839 int todo = len;
|
|
|
2840 int c;
|
|
|
2841 int atstart;
|
|
|
2842 char_u *pat;
|
|
|
2843 char_u *pp;
|
|
|
2844 char_u *cp;
|
|
485
|
2845 char_u *ap;
|
|
483
|
2846
|
|
|
2847 if (todo < 2)
|
|
|
2848 return SP_FORMERROR; /* need at least two bytes */
|
|
|
2849
|
|
|
2850 --todo;
|
|
|
2851 c = getc(fd); /* <compmax> */
|
|
|
2852 if (c < 2)
|
|
|
2853 c = MAXWLEN;
|
|
|
2854 slang->sl_compmax = c;
|
|
|
2855
|
|
|
2856 --todo;
|
|
|
2857 c = getc(fd); /* <compminlen> */
|
|
|
2858 if (c < 1)
|
|
|
2859 c = 3;
|
|
|
2860 slang->sl_compminlen = c;
|
|
|
2861
|
|
|
2862 --todo;
|
|
|
2863 c = getc(fd); /* <compsylmax> */
|
|
|
2864 if (c < 1)
|
|
|
2865 c = MAXWLEN;
|
|
|
2866 slang->sl_compsylmax = c;
|
|
|
2867
|
|
|
2868 /* Turn the COMPOUNDFLAGS items into a regexp pattern:
|
|
|
2869 * "a[bc]/a*b+" -> "^\(a[bc]\|a*b\+\)$".
|
|
|
2870 * Inserting backslashes may double the length, "^\(\)$<Nul>" is 7 bytes. */
|
|
|
2871 pat = alloc((unsigned)todo * 2 + 7);
|
|
|
2872 if (pat == NULL)
|
|
|
2873 return SP_ERROR;
|
|
|
2874
|
|
485
|
2875 /* We also need a list of all flags that can appear at the start and one
|
|
|
2876 * for all flags. */
|
|
483
|
2877 cp = alloc(todo + 1);
|
|
|
2878 if (cp == NULL)
|
|
|
2879 {
|
|
|
2880 vim_free(pat);
|
|
|
2881 return SP_ERROR;
|
|
|
2882 }
|
|
|
2883 slang->sl_compstartflags = cp;
|
|
|
2884 *cp = NUL;
|
|
|
2885
|
|
485
|
2886 ap = alloc(todo + 1);
|
|
|
2887 if (ap == NULL)
|
|
|
2888 {
|
|
|
2889 vim_free(pat);
|
|
|
2890 return SP_ERROR;
|
|
|
2891 }
|
|
|
2892 slang->sl_compallflags = ap;
|
|
|
2893 *ap = NUL;
|
|
|
2894
|
|
483
|
2895 pp = pat;
|
|
|
2896 *pp++ = '^';
|
|
|
2897 *pp++ = '\\';
|
|
|
2898 *pp++ = '(';
|
|
|
2899
|
|
|
2900 atstart = 1;
|
|
|
2901 while (todo-- > 0)
|
|
|
2902 {
|
|
|
2903 c = getc(fd); /* <compflags> */
|
|
485
|
2904
|
|
|
2905 /* Add all flags to "sl_compallflags". */
|
|
|
2906 if (vim_strchr((char_u *)"+*[]/", c) == NULL
|
|
|
2907 && vim_strchr(slang->sl_compallflags, c) == NULL)
|
|
|
2908 {
|
|
|
2909 *ap++ = c;
|
|
|
2910 *ap = NUL;
|
|
|
2911 }
|
|
|
2912
|
|
483
|
2913 if (atstart != 0)
|
|
|
2914 {
|
|
|
2915 /* At start of item: copy flags to "sl_compstartflags". For a
|
|
|
2916 * [abc] item set "atstart" to 2 and copy up to the ']'. */
|
|
|
2917 if (c == '[')
|
|
|
2918 atstart = 2;
|
|
|
2919 else if (c == ']')
|
|
|
2920 atstart = 0;
|
|
|
2921 else
|
|
|
2922 {
|
|
|
2923 if (vim_strchr(slang->sl_compstartflags, c) == NULL)
|
|
|
2924 {
|
|
|
2925 *cp++ = c;
|
|
|
2926 *cp = NUL;
|
|
|
2927 }
|
|
|
2928 if (atstart == 1)
|
|
|
2929 atstart = 0;
|
|
|
2930 }
|
|
|
2931 }
|
|
|
2932 if (c == '/') /* slash separates two items */
|
|
|
2933 {
|
|
|
2934 *pp++ = '\\';
|
|
|
2935 *pp++ = '|';
|
|
|
2936 atstart = 1;
|
|
|
2937 }
|
|
|
2938 else /* normal char, "[abc]" and '*' are copied as-is */
|
|
|
2939 {
|
|
|
2940 if (c == '+')
|
|
|
2941 *pp++ = '\\'; /* "a+" becomes "a\+" */
|
|
|
2942 *pp++ = c;
|
|
|
2943 }
|
|
|
2944 }
|
|
|
2945
|
|
|
2946 *pp++ = '\\';
|
|
|
2947 *pp++ = ')';
|
|
|
2948 *pp++ = '$';
|
|
|
2949 *pp = NUL;
|
|
|
2950
|
|
|
2951 slang->sl_compprog = vim_regcomp(pat, RE_MAGIC + RE_STRING + RE_STRICT);
|
|
|
2952 vim_free(pat);
|
|
|
2953 if (slang->sl_compprog == NULL)
|
|
|
2954 return SP_FORMERROR;
|
|
|
2955
|
|
|
2956 return 0;
|
|
|
2957 }
|
|
|
2958
|
|
|
2959 #define SY_MAXLEN 30
|
|
|
2960 typedef struct syl_item_S
|
|
|
2961 {
|
|
|
2962 char_u sy_chars[SY_MAXLEN]; /* the sequence of chars */
|
|
|
2963 int sy_len;
|
|
|
2964 } syl_item_T;
|
|
|
2965
|
|
|
2966 /*
|
|
|
2967 * Truncate "slang->sl_syllable" at the first slash and put the following items
|
|
|
2968 * in "slang->sl_syl_items".
|
|
|
2969 */
|
|
|
2970 static int
|
|
|
2971 init_syl_tab(slang)
|
|
|
2972 slang_T *slang;
|
|
|
2973 {
|
|
|
2974 char_u *p;
|
|
|
2975 char_u *s;
|
|
|
2976 int l;
|
|
|
2977 syl_item_T *syl;
|
|
|
2978
|
|
|
2979 ga_init2(&slang->sl_syl_items, sizeof(syl_item_T), 4);
|
|
|
2980 p = vim_strchr(slang->sl_syllable, '/');
|
|
|
2981 while (p != NULL)
|
|
|
2982 {
|
|
|
2983 *p++ = NUL;
|
|
|
2984 if (p == NUL)
|
|
|
2985 break;
|
|
|
2986 s = p;
|
|
|
2987 p = vim_strchr(p, '/');
|
|
|
2988 if (p == NULL)
|
|
|
2989 l = STRLEN(s);
|
|
|
2990 else
|
|
|
2991 l = p - s;
|
|
|
2992 if (l >= SY_MAXLEN)
|
|
|
2993 return SP_FORMERROR;
|
|
|
2994 if (ga_grow(&slang->sl_syl_items, 1) == FAIL)
|
|
|
2995 return SP_ERROR;
|
|
|
2996 syl = ((syl_item_T *)slang->sl_syl_items.ga_data)
|
|
|
2997 + slang->sl_syl_items.ga_len++;
|
|
|
2998 vim_strncpy(syl->sy_chars, s, l);
|
|
|
2999 syl->sy_len = l;
|
|
|
3000 }
|
|
|
3001 return OK;
|
|
|
3002 }
|
|
|
3003
|
|
|
3004 /*
|
|
|
3005 * Count the number of syllables in "word".
|
|
|
3006 * When "word" contains spaces the syllables after the last space are counted.
|
|
|
3007 * Returns zero if syllables are not defines.
|
|
|
3008 */
|
|
|
3009 static int
|
|
|
3010 count_syllables(slang, word)
|
|
|
3011 slang_T *slang;
|
|
|
3012 char_u *word;
|
|
|
3013 {
|
|
|
3014 int cnt = 0;
|
|
|
3015 int skip = FALSE;
|
|
|
3016 char_u *p;
|
|
|
3017 int len;
|
|
|
3018 int i;
|
|
|
3019 syl_item_T *syl;
|
|
|
3020 int c;
|
|
|
3021
|
|
|
3022 if (slang->sl_syllable == NULL)
|
|
|
3023 return 0;
|
|
|
3024
|
|
|
3025 for (p = word; *p != NUL; p += len)
|
|
|
3026 {
|
|
|
3027 /* When running into a space reset counter. */
|
|
|
3028 if (*p == ' ')
|
|
|
3029 {
|
|
|
3030 len = 1;
|
|
|
3031 cnt = 0;
|
|
|
3032 continue;
|
|
|
3033 }
|
|
|
3034
|
|
|
3035 /* Find longest match of syllable items. */
|
|
|
3036 len = 0;
|
|
|
3037 for (i = 0; i < slang->sl_syl_items.ga_len; ++i)
|
|
|
3038 {
|
|
|
3039 syl = ((syl_item_T *)slang->sl_syl_items.ga_data) + i;
|
|
|
3040 if (syl->sy_len > len
|
|
|
3041 && STRNCMP(p, syl->sy_chars, syl->sy_len) == 0)
|
|
|
3042 len = syl->sy_len;
|
|
|
3043 }
|
|
|
3044 if (len != 0) /* found a match, count syllable */
|
|
|
3045 {
|
|
|
3046 ++cnt;
|
|
|
3047 skip = FALSE;
|
|
|
3048 }
|
|
|
3049 else
|
|
|
3050 {
|
|
|
3051 /* No recognized syllable item, at least a syllable char then? */
|
|
|
3052 #ifdef FEAT_MBYTE
|
|
|
3053 c = mb_ptr2char(p);
|
|
|
3054 len = (*mb_ptr2len)(p);
|
|
|
3055 #else
|
|
|
3056 c = *p;
|
|
|
3057 len = 1;
|
|
|
3058 #endif
|
|
|
3059 if (vim_strchr(slang->sl_syllable, c) == NULL)
|
|
|
3060 skip = FALSE; /* No, search for next syllable */
|
|
|
3061 else if (!skip)
|
|
|
3062 {
|
|
|
3063 ++cnt; /* Yes, count it */
|
|
|
3064 skip = TRUE; /* don't count following syllable chars */
|
|
|
3065 }
|
|
|
3066 }
|
|
|
3067 }
|
|
|
3068 return cnt;
|
|
376
|
3069 }
|
|
|
3070
|
|
381
|
3071 /*
|
|
|
3072 * Set the SOFOFROM and SOFOTO items in language "lp".
|
|
483
|
3073 * Returns SP_*ERROR flags when there is something wrong.
|
|
381
|
3074 */
|
|
|
3075 static int
|
|
|
3076 set_sofo(lp, from, to)
|
|
|
3077 slang_T *lp;
|
|
|
3078 char_u *from;
|
|
|
3079 char_u *to;
|
|
|
3080 {
|
|
|
3081 int i;
|
|
|
3082
|
|
|
3083 #ifdef FEAT_MBYTE
|
|
|
3084 garray_T *gap;
|
|
|
3085 char_u *s;
|
|
|
3086 char_u *p;
|
|
|
3087 int c;
|
|
|
3088 int *inp;
|
|
|
3089
|
|
|
3090 if (has_mbyte)
|
|
|
3091 {
|
|
|
3092 /* Use "sl_sal" as an array with 256 pointers to a list of wide
|
|
|
3093 * characters. The index is the low byte of the character.
|
|
|
3094 * The list contains from-to pairs with a terminating NUL.
|
|
|
3095 * sl_sal_first[] is used for latin1 "from" characters. */
|
|
|
3096 gap = &lp->sl_sal;
|
|
|
3097 ga_init2(gap, sizeof(int *), 1);
|
|
|
3098 if (ga_grow(gap, 256) == FAIL)
|
|
483
|
3099 return SP_ERROR;
|
|
381
|
3100 vim_memset(gap->ga_data, 0, sizeof(int *) * 256);
|
|
|
3101 gap->ga_len = 256;
|
|
|
3102
|
|
|
3103 /* First count the number of items for each list. Temporarily use
|
|
|
3104 * sl_sal_first[] for this. */
|
|
|
3105 for (p = from, s = to; *p != NUL && *s != NUL; )
|
|
|
3106 {
|
|
474
|
3107 c = mb_cptr2char_adv(&p);
|
|
|
3108 mb_cptr_adv(s);
|
|
381
|
3109 if (c >= 256)
|
|
|
3110 ++lp->sl_sal_first[c & 0xff];
|
|
|
3111 }
|
|
|
3112 if (*p != NUL || *s != NUL) /* lengths differ */
|
|
483
|
3113 return SP_FORMERROR;
|
|
381
|
3114
|
|
|
3115 /* Allocate the lists. */
|
|
|
3116 for (i = 0; i < 256; ++i)
|
|
|
3117 if (lp->sl_sal_first[i] > 0)
|
|
|
3118 {
|
|
|
3119 p = alloc(sizeof(int) * (lp->sl_sal_first[i] * 2 + 1));
|
|
|
3120 if (p == NULL)
|
|
483
|
3121 return SP_ERROR;
|
|
381
|
3122 ((int **)gap->ga_data)[i] = (int *)p;
|
|
|
3123 *(int *)p = 0;
|
|
|
3124 }
|
|
|
3125
|
|
|
3126 /* Put the characters up to 255 in sl_sal_first[] the rest in a sl_sal
|
|
|
3127 * list. */
|
|
|
3128 vim_memset(lp->sl_sal_first, 0, sizeof(salfirst_T) * 256);
|
|
|
3129 for (p = from, s = to; *p != NUL && *s != NUL; )
|
|
|
3130 {
|
|
474
|
3131 c = mb_cptr2char_adv(&p);
|
|
|
3132 i = mb_cptr2char_adv(&s);
|
|
381
|
3133 if (c >= 256)
|
|
|
3134 {
|
|
|
3135 /* Append the from-to chars at the end of the list with
|
|
|
3136 * the low byte. */
|
|
|
3137 inp = ((int **)gap->ga_data)[c & 0xff];
|
|
|
3138 while (*inp != 0)
|
|
|
3139 ++inp;
|
|
|
3140 *inp++ = c; /* from char */
|
|
|
3141 *inp++ = i; /* to char */
|
|
|
3142 *inp++ = NUL; /* NUL at the end */
|
|
|
3143 }
|
|
|
3144 else
|
|
|
3145 /* mapping byte to char is done in sl_sal_first[] */
|
|
|
3146 lp->sl_sal_first[c] = i;
|
|
|
3147 }
|
|
|
3148 }
|
|
|
3149 else
|
|
|
3150 #endif
|
|
|
3151 {
|
|
|
3152 /* mapping bytes to bytes is done in sl_sal_first[] */
|
|
|
3153 if (STRLEN(from) != STRLEN(to))
|
|
483
|
3154 return SP_FORMERROR;
|
|
381
|
3155
|
|
|
3156 for (i = 0; to[i] != NUL; ++i)
|
|
|
3157 lp->sl_sal_first[from[i]] = to[i];
|
|
|
3158 lp->sl_sal.ga_len = 1; /* indicates we have soundfolding */
|
|
|
3159 }
|
|
|
3160
|
|
483
|
3161 return 0;
|
|
381
|
3162 }
|
|
|
3163
|
|
|
3164 /*
|
|
|
3165 * Fill the first-index table for "lp".
|
|
|
3166 */
|
|
|
3167 static void
|
|
|
3168 set_sal_first(lp)
|
|
|
3169 slang_T *lp;
|
|
|
3170 {
|
|
|
3171 salfirst_T *sfirst;
|
|
|
3172 int i;
|
|
|
3173 salitem_T *smp;
|
|
|
3174 int c;
|
|
|
3175 garray_T *gap = &lp->sl_sal;
|
|
|
3176
|
|
|
3177 sfirst = lp->sl_sal_first;
|
|
|
3178 for (i = 0; i < 256; ++i)
|
|
|
3179 sfirst[i] = -1;
|
|
|
3180 smp = (salitem_T *)gap->ga_data;
|
|
|
3181 for (i = 0; i < gap->ga_len; ++i)
|
|
|
3182 {
|
|
|
3183 #ifdef FEAT_MBYTE
|
|
|
3184 if (has_mbyte)
|
|
|
3185 /* Use the lowest byte of the first character. For latin1 it's
|
|
|
3186 * the character, for other encodings it should differ for most
|
|
|
3187 * characters. */
|
|
|
3188 c = *smp[i].sm_lead_w & 0xff;
|
|
|
3189 else
|
|
|
3190 #endif
|
|
|
3191 c = *smp[i].sm_lead;
|
|
|
3192 if (sfirst[c] == -1)
|
|
|
3193 {
|
|
|
3194 sfirst[c] = i;
|
|
|
3195 #ifdef FEAT_MBYTE
|
|
|
3196 if (has_mbyte)
|
|
|
3197 {
|
|
|
3198 int n;
|
|
|
3199
|
|
|
3200 /* Make sure all entries with this byte are following each
|
|
|
3201 * other. Move the ones that are in the wrong position. Do
|
|
|
3202 * keep the same ordering! */
|
|
|
3203 while (i + 1 < gap->ga_len
|
|
|
3204 && (*smp[i + 1].sm_lead_w & 0xff) == c)
|
|
|
3205 /* Skip over entry with same index byte. */
|
|
|
3206 ++i;
|
|
|
3207
|
|
|
3208 for (n = 1; i + n < gap->ga_len; ++n)
|
|
|
3209 if ((*smp[i + n].sm_lead_w & 0xff) == c)
|
|
|
3210 {
|
|
|
3211 salitem_T tsal;
|
|
|
3212
|
|
|
3213 /* Move entry with same index byte after the entries
|
|
|
3214 * we already found. */
|
|
|
3215 ++i;
|
|
|
3216 --n;
|
|
|
3217 tsal = smp[i + n];
|
|
|
3218 mch_memmove(smp + i + 1, smp + i,
|
|
|
3219 sizeof(salitem_T) * n);
|
|
|
3220 smp[i] = tsal;
|
|
|
3221 }
|
|
|
3222 }
|
|
|
3223 #endif
|
|
|
3224 }
|
|
|
3225 }
|
|
|
3226 }
|
|
376
|
3227
|
|
372
|
3228 #ifdef FEAT_MBYTE
|
|
|
3229 /*
|
|
|
3230 * Turn a multi-byte string into a wide character string.
|
|
|
3231 * Return it in allocated memory (NULL for out-of-memory)
|
|
|
3232 */
|
|
|
3233 static int *
|
|
|
3234 mb_str2wide(s)
|
|
|
3235 char_u *s;
|
|
|
3236 {
|
|
|
3237 int *res;
|
|
|
3238 char_u *p;
|
|
|
3239 int i = 0;
|
|
|
3240
|
|
|
3241 res = (int *)alloc(sizeof(int) * (mb_charlen(s) + 1));
|
|
|
3242 if (res != NULL)
|
|
|
3243 {
|
|
|
3244 for (p = s; *p != NUL; )
|
|
|
3245 res[i++] = mb_ptr2char_adv(&p);
|
|
|
3246 res[i] = NUL;
|
|
|
3247 }
|
|
|
3248 return res;
|
|
|
3249 }
|
|
|
3250 #endif
|
|
|
3251
|
|
236
|
3252 /*
|
|
300
|
3253 * Read one row of siblings from the spell file and store it in the byte array
|
|
|
3254 * "byts" and index array "idxs". Recursively read the children.
|
|
|
3255 *
|
|
346
|
3256 * NOTE: The code here must match put_node().
|
|
300
|
3257 *
|
|
|
3258 * Returns the index follosing the siblings.
|
|
|
3259 * Returns -1 if the file is shorter than expected.
|
|
|
3260 * Returns -2 if there is a format error.
|
|
236
|
3261 */
|
|
324
|
3262 static idx_T
|
|
339
|
3263 read_tree(fd, byts, idxs, maxidx, startidx, prefixtree, maxprefcondnr)
|
|
300
|
3264 FILE *fd;
|
|
|
3265 char_u *byts;
|
|
324
|
3266 idx_T *idxs;
|
|
300
|
3267 int maxidx; /* size of arrays */
|
|
324
|
3268 idx_T startidx; /* current index in "byts" and "idxs" */
|
|
339
|
3269 int prefixtree; /* TRUE for reading PREFIXTREE */
|
|
|
3270 int maxprefcondnr; /* maximum for <prefcondnr> */
|
|
236
|
3271 {
|
|
300
|
3272 int len;
|
|
|
3273 int i;
|
|
|
3274 int n;
|
|
324
|
3275 idx_T idx = startidx;
|
|
300
|
3276 int c;
|
|
366
|
3277 int c2;
|
|
300
|
3278 #define SHARED_MASK 0x8000000
|
|
236
|
3279
|
|
300
|
3280 len = getc(fd); /* <siblingcount> */
|
|
|
3281 if (len <= 0)
|
|
|
3282 return -1;
|
|
|
3283
|
|
|
3284 if (startidx + len >= maxidx)
|
|
|
3285 return -2;
|
|
|
3286 byts[idx++] = len;
|
|
|
3287
|
|
|
3288 /* Read the byte values, flag/region bytes and shared indexes. */
|
|
|
3289 for (i = 1; i <= len; ++i)
|
|
236
|
3290 {
|
|
300
|
3291 c = getc(fd); /* <byte> */
|
|
|
3292 if (c < 0)
|
|
|
3293 return -1;
|
|
|
3294 if (c <= BY_SPECIAL)
|
|
|
3295 {
|
|
366
|
3296 if (c == BY_NOFLAGS && !prefixtree)
|
|
300
|
3297 {
|
|
|
3298 /* No flags, all regions. */
|
|
|
3299 idxs[idx] = 0;
|
|
|
3300 c = 0;
|
|
|
3301 }
|
|
390
|
3302 else if (c != BY_INDEX)
|
|
300
|
3303 {
|
|
339
|
3304 if (prefixtree)
|
|
|
3305 {
|
|
455
|
3306 /* Read the optional pflags byte, the prefix ID and the
|
|
|
3307 * condition nr. In idxs[] store the prefix ID in the low
|
|
|
3308 * byte, the condition index shifted up 8 bits, the flags
|
|
|
3309 * shifted up 24 bits. */
|
|
|
3310 if (c == BY_FLAGS)
|
|
|
3311 c = getc(fd) << 24; /* <pflags> */
|
|
|
3312 else
|
|
|
3313 c = 0;
|
|
|
3314
|
|
481
|
3315 c |= getc(fd); /* <affixID> */
|
|
455
|
3316
|
|
339
|
3317 n = (getc(fd) << 8) + getc(fd); /* <prefcondnr> */
|
|
|
3318 if (n >= maxprefcondnr)
|
|
|
3319 return -2;
|
|
455
|
3320 c |= (n << 8);
|
|
339
|
3321 }
|
|
390
|
3322 else /* c must be BY_FLAGS or BY_FLAGS2 */
|
|
339
|
3323 {
|
|
|
3324 /* Read flags and optional region and prefix ID. In
|
|
390
|
3325 * idxs[] the flags go in the low two bytes, region above
|
|
|
3326 * that and prefix ID above the region. */
|
|
|
3327 c2 = c;
|
|
339
|
3328 c = getc(fd); /* <flags> */
|
|
390
|
3329 if (c2 == BY_FLAGS2)
|
|
|
3330 c = (getc(fd) << 8) + c; /* <flags2> */
|
|
339
|
3331 if (c & WF_REGION)
|
|
390
|
3332 c = (getc(fd) << 16) + c; /* <region> */
|
|
481
|
3333 if (c & WF_AFX)
|
|
|
3334 c = (getc(fd) << 24) + c; /* <affixID> */
|
|
339
|
3335 }
|
|
|
3336
|
|
300
|
3337 idxs[idx] = c;
|
|
|
3338 c = 0;
|
|
|
3339 }
|
|
|
3340 else /* c == BY_INDEX */
|
|
|
3341 {
|
|
|
3342 /* <nodeidx> */
|
|
|
3343 n = (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
3344 if (n < 0 || n >= maxidx)
|
|
|
3345 return -2;
|
|
|
3346 idxs[idx] = n + SHARED_MASK;
|
|
|
3347 c = getc(fd); /* <xbyte> */
|
|
|
3348 }
|
|
|
3349 }
|
|
|
3350 byts[idx++] = c;
|
|
236
|
3351 }
|
|
|
3352
|
|
300
|
3353 /* Recursively read the children for non-shared siblings.
|
|
|
3354 * Skip the end-of-word ones (zero byte value) and the shared ones (and
|
|
|
3355 * remove SHARED_MASK) */
|
|
|
3356 for (i = 1; i <= len; ++i)
|
|
|
3357 if (byts[startidx + i] != 0)
|
|
|
3358 {
|
|
|
3359 if (idxs[startidx + i] & SHARED_MASK)
|
|
|
3360 idxs[startidx + i] &= ~SHARED_MASK;
|
|
|
3361 else
|
|
|
3362 {
|
|
|
3363 idxs[startidx + i] = idx;
|
|
339
|
3364 idx = read_tree(fd, byts, idxs, maxidx, idx,
|
|
|
3365 prefixtree, maxprefcondnr);
|
|
300
|
3366 if (idx < 0)
|
|
|
3367 break;
|
|
|
3368 }
|
|
|
3369 }
|
|
236
|
3370
|
|
300
|
3371 return idx;
|
|
236
|
3372 }
|
|
|
3373
|
|
|
3374 /*
|
|
|
3375 * Parse 'spelllang' and set buf->b_langp accordingly.
|
|
351
|
3376 * Returns NULL if it's OK, an error message otherwise.
|
|
236
|
3377 */
|
|
|
3378 char_u *
|
|
|
3379 did_set_spelllang(buf)
|
|
|
3380 buf_T *buf;
|
|
|
3381 {
|
|
|
3382 garray_T ga;
|
|
351
|
3383 char_u *splp;
|
|
236
|
3384 char_u *region;
|
|
416
|
3385 char_u region_cp[3];
|
|
355
|
3386 int filename;
|
|
236
|
3387 int region_mask;
|
|
|
3388 slang_T *lp;
|
|
|
3389 int c;
|
|
351
|
3390 char_u lang[MAXWLEN + 1];
|
|
323
|
3391 char_u spf_name[MAXPATHL];
|
|
351
|
3392 int len;
|
|
|
3393 char_u *p;
|
|
381
|
3394 int round;
|
|
385
|
3395 char_u *spf;
|
|
389
|
3396 char_u *use_region = NULL;
|
|
|
3397 int dont_use_region = FALSE;
|
|
236
|
3398
|
|
|
3399 ga_init2(&ga, sizeof(langp_T), 2);
|
|
376
|
3400 clear_midword(buf);
|
|
236
|
3401
|
|
351
|
3402 /* loop over comma separated language names. */
|
|
|
3403 for (splp = buf->b_p_spl; *splp != NUL; )
|
|
|
3404 {
|
|
|
3405 /* Get one language name. */
|
|
|
3406 copy_option_part(&splp, lang, MAXWLEN, ",");
|
|
|
3407
|
|
240
|
3408 region = NULL;
|
|
351
|
3409 len = STRLEN(lang);
|
|
355
|
3410
|
|
|
3411 /* If the name ends in ".spl" use it as the name of the spell file.
|
|
|
3412 * If there is a region name let "region" point to it and remove it
|
|
|
3413 * from the name. */
|
|
|
3414 if (len > 4 && fnamecmp(lang + len - 4, ".spl") == 0)
|
|
236
|
3415 {
|
|
355
|
3416 filename = TRUE;
|
|
|
3417
|
|
416
|
3418 /* Locate a region and remove it from the file name. */
|
|
|
3419 p = vim_strchr(gettail(lang), '_');
|
|
|
3420 if (p != NULL && ASCII_ISALPHA(p[1]) && ASCII_ISALPHA(p[2])
|
|
|
3421 && !ASCII_ISALPHA(p[3]))
|
|
|
3422 {
|
|
|
3423 vim_strncpy(region_cp, p + 1, 2);
|
|
|
3424 mch_memmove(p, p + 3, len - (p - lang) - 2);
|
|
|
3425 len -= 3;
|
|
|
3426 region = region_cp;
|
|
|
3427 }
|
|
|
3428 else
|
|
|
3429 dont_use_region = TRUE;
|
|
|
3430
|
|
355
|
3431 /* Check if we loaded this language before. */
|
|
|
3432 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
3433 if (fullpathcmp(lang, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
3434 break;
|
|
236
|
3435 }
|
|
355
|
3436 else
|
|
|
3437 {
|
|
|
3438 filename = FALSE;
|
|
|
3439 if (len > 3 && lang[len - 3] == '_')
|
|
|
3440 {
|
|
|
3441 region = lang + len - 2;
|
|
|
3442 len -= 3;
|
|
|
3443 lang[len] = NUL;
|
|
389
|
3444 }
|
|
|
3445 else
|
|
|
3446 dont_use_region = TRUE;
|
|
355
|
3447
|
|
|
3448 /* Check if we loaded this language before. */
|
|
|
3449 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
3450 if (STRICMP(lang, lp->sl_name) == 0)
|
|
|
3451 break;
|
|
|
3452 }
|
|
236
|
3453
|
|
416
|
3454 if (region != NULL)
|
|
|
3455 {
|
|
|
3456 /* If the region differs from what was used before then don't
|
|
|
3457 * use it for 'spellfile'. */
|
|
|
3458 if (use_region != NULL && STRCMP(region, use_region) != 0)
|
|
|
3459 dont_use_region = TRUE;
|
|
|
3460 use_region = region;
|
|
|
3461 }
|
|
|
3462
|
|
351
|
3463 /* If not found try loading the language now. */
|
|
236
|
3464 if (lp == NULL)
|
|
355
|
3465 {
|
|
|
3466 if (filename)
|
|
|
3467 (void)spell_load_file(lang, lang, NULL, FALSE);
|
|
|
3468 else
|
|
|
3469 spell_load_lang(lang);
|
|
|
3470 }
|
|
236
|
3471
|
|
307
|
3472 /*
|
|
351
|
3473 * Loop over the languages, there can be several files for "lang".
|
|
307
|
3474 */
|
|
|
3475 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
355
|
3476 if (filename ? fullpathcmp(lang, lp->sl_fname, FALSE) == FPC_SAME
|
|
|
3477 : STRICMP(lang, lp->sl_name) == 0)
|
|
236
|
3478 {
|
|
316
|
3479 region_mask = REGION_ALL;
|
|
355
|
3480 if (!filename && region != NULL)
|
|
236
|
3481 {
|
|
307
|
3482 /* find region in sl_regions */
|
|
|
3483 c = find_region(lp->sl_regions, region);
|
|
|
3484 if (c == REGION_ALL)
|
|
|
3485 {
|
|
389
|
3486 if (lp->sl_add)
|
|
|
3487 {
|
|
|
3488 if (*lp->sl_regions != NUL)
|
|
|
3489 /* This addition file is for other regions. */
|
|
|
3490 region_mask = 0;
|
|
|
3491 }
|
|
|
3492 else
|
|
|
3493 /* This is probably an error. Give a warning and
|
|
|
3494 * accept the words anyway. */
|
|
351
|
3495 smsg((char_u *)
|
|
|
3496 _("Warning: region %s not supported"),
|
|
|
3497 region);
|
|
307
|
3498 }
|
|
|
3499 else
|
|
|
3500 region_mask = 1 << c;
|
|
236
|
3501 }
|
|
307
|
3502
|
|
389
|
3503 if (region_mask != 0)
|
|
307
|
3504 {
|
|
389
|
3505 if (ga_grow(&ga, 1) == FAIL)
|
|
|
3506 {
|
|
|
3507 ga_clear(&ga);
|
|
|
3508 return e_outofmem;
|
|
|
3509 }
|
|
|
3510 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = lp;
|
|
|
3511 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask;
|
|
|
3512 ++ga.ga_len;
|
|
|
3513 use_midword(lp, buf);
|
|
307
|
3514 }
|
|
385
|
3515 }
|
|
|
3516 }
|
|
|
3517
|
|
|
3518 /* round 0: load int_wordlist, if possible.
|
|
|
3519 * round 1: load first name in 'spellfile'.
|
|
|
3520 * round 2: load second name in 'spellfile.
|
|
|
3521 * etc. */
|
|
|
3522 spf = curbuf->b_p_spf;
|
|
|
3523 for (round = 0; round == 0 || *spf != NUL; ++round)
|
|
|
3524 {
|
|
|
3525 if (round == 0)
|
|
|
3526 {
|
|
|
3527 /* Internal wordlist, if there is one. */
|
|
|
3528 if (int_wordlist == NULL)
|
|
381
|
3529 continue;
|
|
385
|
3530 int_wordlist_spl(spf_name);
|
|
381
|
3531 }
|
|
|
3532 else
|
|
|
3533 {
|
|
385
|
3534 /* One entry in 'spellfile'. */
|
|
|
3535 copy_option_part(&spf, spf_name, MAXPATHL - 5, ",");
|
|
|
3536 STRCAT(spf_name, ".spl");
|
|
|
3537
|
|
|
3538 /* If it was already found above then skip it. */
|
|
|
3539 for (c = 0; c < ga.ga_len; ++c)
|
|
|
3540 if (fullpathcmp(spf_name,
|
|
|
3541 LANGP_ENTRY(ga, c)->lp_slang->sl_fname,
|
|
|
3542 FALSE) == FPC_SAME)
|
|
|
3543 break;
|
|
|
3544 if (c < ga.ga_len)
|
|
381
|
3545 continue;
|
|
|
3546 }
|
|
|
3547
|
|
351
|
3548 /* Check if it was loaded already. */
|
|
323
|
3549 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
3550 if (fullpathcmp(spf_name, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
3551 break;
|
|
|
3552 if (lp == NULL)
|
|
|
3553 {
|
|
351
|
3554 /* Not loaded, try loading it now. The language name includes the
|
|
385
|
3555 * region name, the region is ignored otherwise. for int_wordlist
|
|
|
3556 * use an arbitrary name. */
|
|
|
3557 if (round == 0)
|
|
|
3558 STRCPY(lang, "internal wordlist");
|
|
|
3559 else
|
|
|
3560 {
|
|
|
3561 vim_strncpy(lang, gettail(spf_name), MAXWLEN);
|
|
381
|
3562 p = vim_strchr(lang, '.');
|
|
|
3563 if (p != NULL)
|
|
|
3564 *p = NUL; /* truncate at ".encoding.add" */
|
|
|
3565 }
|
|
351
|
3566 lp = spell_load_file(spf_name, lang, NULL, TRUE);
|
|
323
|
3567 }
|
|
|
3568 if (lp != NULL && ga_grow(&ga, 1) == OK)
|
|
|
3569 {
|
|
389
|
3570 region_mask = REGION_ALL;
|
|
|
3571 if (use_region != NULL && !dont_use_region)
|
|
|
3572 {
|
|
|
3573 /* find region in sl_regions */
|
|
|
3574 c = find_region(lp->sl_regions, use_region);
|
|
|
3575 if (c != REGION_ALL)
|
|
|
3576 region_mask = 1 << c;
|
|
|
3577 else if (*lp->sl_regions != NUL)
|
|
|
3578 /* This spell file is for other regions. */
|
|
|
3579 region_mask = 0;
|
|
|
3580 }
|
|
|
3581
|
|
|
3582 if (region_mask != 0)
|
|
|
3583 {
|
|
|
3584 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = lp;
|
|
|
3585 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask;
|
|
|
3586 ++ga.ga_len;
|
|
|
3587 use_midword(lp, buf);
|
|
|
3588 }
|
|
323
|
3589 }
|
|
|
3590 }
|
|
|
3591
|
|
236
|
3592 /* Add a NULL entry to mark the end of the list. */
|
|
|
3593 if (ga_grow(&ga, 1) == FAIL)
|
|
|
3594 {
|
|
|
3595 ga_clear(&ga);
|
|
|
3596 return e_outofmem;
|
|
|
3597 }
|
|
|
3598 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = NULL;
|
|
|
3599 ++ga.ga_len;
|
|
|
3600
|
|
|
3601 /* Everything is fine, store the new b_langp value. */
|
|
|
3602 ga_clear(&buf->b_langp);
|
|
|
3603 buf->b_langp = ga;
|
|
|
3604
|
|
|
3605 return NULL;
|
|
|
3606 }
|
|
|
3607
|
|
|
3608 /*
|
|
376
|
3609 * Clear the midword characters for buffer "buf".
|
|
|
3610 */
|
|
|
3611 static void
|
|
|
3612 clear_midword(buf)
|
|
|
3613 buf_T *buf;
|
|
|
3614 {
|
|
|
3615 vim_memset(buf->b_spell_ismw, 0, 256);
|
|
|
3616 #ifdef FEAT_MBYTE
|
|
|
3617 vim_free(buf->b_spell_ismw_mb);
|
|
|
3618 buf->b_spell_ismw_mb = NULL;
|
|
|
3619 #endif
|
|
|
3620 }
|
|
|
3621
|
|
|
3622 /*
|
|
|
3623 * Use the "sl_midword" field of language "lp" for buffer "buf".
|
|
|
3624 * They add up to any currently used midword characters.
|
|
|
3625 */
|
|
|
3626 static void
|
|
|
3627 use_midword(lp, buf)
|
|
|
3628 slang_T *lp;
|
|
|
3629 buf_T *buf;
|
|
|
3630 {
|
|
|
3631 char_u *p;
|
|
|
3632
|
|
389
|
3633 if (lp->sl_midword == NULL) /* there aren't any */
|
|
|
3634 return;
|
|
|
3635
|
|
376
|
3636 for (p = lp->sl_midword; *p != NUL; )
|
|
|
3637 #ifdef FEAT_MBYTE
|
|
|
3638 if (has_mbyte)
|
|
|
3639 {
|
|
|
3640 int c, l, n;
|
|
|
3641 char_u *bp;
|
|
|
3642
|
|
|
3643 c = mb_ptr2char(p);
|
|
474
|
3644 l = (*mb_ptr2len)(p);
|
|
|
3645 if (c < 256 && l <= 2)
|
|
376
|
3646 buf->b_spell_ismw[c] = TRUE;
|
|
|
3647 else if (buf->b_spell_ismw_mb == NULL)
|
|
|
3648 /* First multi-byte char in "b_spell_ismw_mb". */
|
|
|
3649 buf->b_spell_ismw_mb = vim_strnsave(p, l);
|
|
|
3650 else
|
|
|
3651 {
|
|
|
3652 /* Append multi-byte chars to "b_spell_ismw_mb". */
|
|
|
3653 n = STRLEN(buf->b_spell_ismw_mb);
|
|
|
3654 bp = vim_strnsave(buf->b_spell_ismw_mb, n + l);
|
|
|
3655 if (bp != NULL)
|
|
|
3656 {
|
|
|
3657 vim_free(buf->b_spell_ismw_mb);
|
|
|
3658 buf->b_spell_ismw_mb = bp;
|
|
|
3659 vim_strncpy(bp + n, p, l);
|
|
|
3660 }
|
|
|
3661 }
|
|
|
3662 p += l;
|
|
|
3663 }
|
|
|
3664 else
|
|
|
3665 #endif
|
|
|
3666 buf->b_spell_ismw[*p++] = TRUE;
|
|
|
3667 }
|
|
|
3668
|
|
|
3669 /*
|
|
236
|
3670 * Find the region "region[2]" in "rp" (points to "sl_regions").
|
|
|
3671 * Each region is simply stored as the two characters of it's name.
|
|
381
|
3672 * Returns the index if found (first is 0), REGION_ALL if not found.
|
|
236
|
3673 */
|
|
|
3674 static int
|
|
|
3675 find_region(rp, region)
|
|
|
3676 char_u *rp;
|
|
|
3677 char_u *region;
|
|
|
3678 {
|
|
|
3679 int i;
|
|
|
3680
|
|
|
3681 for (i = 0; ; i += 2)
|
|
|
3682 {
|
|
|
3683 if (rp[i] == NUL)
|
|
|
3684 return REGION_ALL;
|
|
|
3685 if (rp[i] == region[0] && rp[i + 1] == region[1])
|
|
|
3686 break;
|
|
|
3687 }
|
|
|
3688 return i / 2;
|
|
|
3689 }
|
|
|
3690
|
|
|
3691 /*
|
|
323
|
3692 * Return case type of word:
|
|
236
|
3693 * w word 0
|
|
300
|
3694 * Word WF_ONECAP
|
|
|
3695 * W WORD WF_ALLCAP
|
|
|
3696 * WoRd wOrd WF_KEEPCAP
|
|
236
|
3697 */
|
|
|
3698 static int
|
|
|
3699 captype(word, end)
|
|
|
3700 char_u *word;
|
|
323
|
3701 char_u *end; /* When NULL use up to NUL byte. */
|
|
236
|
3702 {
|
|
|
3703 char_u *p;
|
|
|
3704 int c;
|
|
|
3705 int firstcap;
|
|
|
3706 int allcap;
|
|
|
3707 int past_second = FALSE; /* past second word char */
|
|
|
3708
|
|
|
3709 /* find first letter */
|
|
376
|
3710 for (p = word; !spell_iswordp_nmw(p); mb_ptr_adv(p))
|
|
323
|
3711 if (end == NULL ? *p == NUL : p >= end)
|
|
236
|
3712 return 0; /* only non-word characters, illegal word */
|
|
|
3713 #ifdef FEAT_MBYTE
|
|
310
|
3714 if (has_mbyte)
|
|
|
3715 c = mb_ptr2char_adv(&p);
|
|
|
3716 else
|
|
236
|
3717 #endif
|
|
310
|
3718 c = *p++;
|
|
324
|
3719 firstcap = allcap = SPELL_ISUPPER(c);
|
|
236
|
3720
|
|
|
3721 /*
|
|
|
3722 * Need to check all letters to find a word with mixed upper/lower.
|
|
|
3723 * But a word with an upper char only at start is a ONECAP.
|
|
|
3724 */
|
|
323
|
3725 for ( ; end == NULL ? *p != NUL : p < end; mb_ptr_adv(p))
|
|
376
|
3726 if (spell_iswordp_nmw(p))
|
|
236
|
3727 {
|
|
455
|
3728 c = PTR2CHAR(p);
|
|
324
|
3729 if (!SPELL_ISUPPER(c))
|
|
236
|
3730 {
|
|
|
3731 /* UUl -> KEEPCAP */
|
|
|
3732 if (past_second && allcap)
|
|
300
|
3733 return WF_KEEPCAP;
|
|
236
|
3734 allcap = FALSE;
|
|
|
3735 }
|
|
|
3736 else if (!allcap)
|
|
|
3737 /* UlU -> KEEPCAP */
|
|
300
|
3738 return WF_KEEPCAP;
|
|
236
|
3739 past_second = TRUE;
|
|
|
3740 }
|
|
|
3741
|
|
|
3742 if (allcap)
|
|
300
|
3743 return WF_ALLCAP;
|
|
236
|
3744 if (firstcap)
|
|
300
|
3745 return WF_ONECAP;
|
|
236
|
3746 return 0;
|
|
|
3747 }
|
|
|
3748
|
|
474
|
3749 /*
|
|
|
3750 * Like captype() but for a KEEPCAP word add ONECAP if the word starts with a
|
|
|
3751 * capital. So that make_case_word() can turn WOrd into Word.
|
|
|
3752 * Add ALLCAP for "WOrD".
|
|
|
3753 */
|
|
|
3754 static int
|
|
|
3755 badword_captype(word, end)
|
|
|
3756 char_u *word;
|
|
|
3757 char_u *end;
|
|
|
3758 {
|
|
|
3759 int flags = captype(word, end);
|
|
475
|
3760 int c;
|
|
474
|
3761 int l, u;
|
|
|
3762 int first;
|
|
|
3763 char_u *p;
|
|
|
3764
|
|
|
3765 if (flags & WF_KEEPCAP)
|
|
|
3766 {
|
|
|
3767 /* Count the number of UPPER and lower case letters. */
|
|
|
3768 l = u = 0;
|
|
|
3769 first = FALSE;
|
|
|
3770 for (p = word; p < end; mb_ptr_adv(p))
|
|
|
3771 {
|
|
475
|
3772 c = PTR2CHAR(p);
|
|
|
3773 if (SPELL_ISUPPER(c))
|
|
474
|
3774 {
|
|
|
3775 ++u;
|
|
|
3776 if (p == word)
|
|
|
3777 first = TRUE;
|
|
|
3778 }
|
|
|
3779 else
|
|
|
3780 ++l;
|
|
|
3781 }
|
|
|
3782
|
|
|
3783 /* If there are more UPPER than lower case letters suggest an
|
|
|
3784 * ALLCAP word. Otherwise, if the first letter is UPPER then
|
|
|
3785 * suggest ONECAP. Exception: "ALl" most likely should be "All",
|
|
|
3786 * require three upper case letters. */
|
|
|
3787 if (u > l && u > 2)
|
|
|
3788 flags |= WF_ALLCAP;
|
|
|
3789 else if (first)
|
|
|
3790 flags |= WF_ONECAP;
|
|
|
3791 }
|
|
|
3792 return flags;
|
|
|
3793 }
|
|
|
3794
|
|
355
|
3795 # if defined(FEAT_MBYTE) || defined(EXITFREE) || defined(PROTO)
|
|
|
3796 /*
|
|
|
3797 * Free all languages.
|
|
|
3798 */
|
|
|
3799 void
|
|
|
3800 spell_free_all()
|
|
|
3801 {
|
|
|
3802 slang_T *lp;
|
|
|
3803 buf_T *buf;
|
|
385
|
3804 char_u fname[MAXPATHL];
|
|
355
|
3805
|
|
|
3806 /* Go through all buffers and handle 'spelllang'. */
|
|
|
3807 for (buf = firstbuf; buf != NULL; buf = buf->b_next)
|
|
|
3808 ga_clear(&buf->b_langp);
|
|
|
3809
|
|
|
3810 while (first_lang != NULL)
|
|
|
3811 {
|
|
|
3812 lp = first_lang;
|
|
|
3813 first_lang = lp->sl_next;
|
|
|
3814 slang_free(lp);
|
|
|
3815 }
|
|
366
|
3816
|
|
385
|
3817 if (int_wordlist != NULL)
|
|
|
3818 {
|
|
|
3819 /* Delete the internal wordlist and its .spl file */
|
|
|
3820 mch_remove(int_wordlist);
|
|
|
3821 int_wordlist_spl(fname);
|
|
|
3822 mch_remove(fname);
|
|
|
3823 vim_free(int_wordlist);
|
|
|
3824 int_wordlist = NULL;
|
|
381
|
3825 }
|
|
|
3826
|
|
366
|
3827 init_spell_chartab();
|
|
355
|
3828 }
|
|
|
3829 # endif
|
|
|
3830
|
|
236
|
3831 # if defined(FEAT_MBYTE) || defined(PROTO)
|
|
|
3832 /*
|
|
|
3833 * Clear all spelling tables and reload them.
|
|
307
|
3834 * Used after 'encoding' is set and when ":mkspell" was used.
|
|
236
|
3835 */
|
|
|
3836 void
|
|
|
3837 spell_reload()
|
|
|
3838 {
|
|
|
3839 buf_T *buf;
|
|
316
|
3840 win_T *wp;
|
|
236
|
3841
|
|
358
|
3842 /* Initialize the table for spell_iswordp(). */
|
|
236
|
3843 init_spell_chartab();
|
|
|
3844
|
|
|
3845 /* Unload all allocated memory. */
|
|
355
|
3846 spell_free_all();
|
|
236
|
3847
|
|
|
3848 /* Go through all buffers and handle 'spelllang'. */
|
|
|
3849 for (buf = firstbuf; buf != NULL; buf = buf->b_next)
|
|
|
3850 {
|
|
316
|
3851 /* Only load the wordlists when 'spelllang' is set and there is a
|
|
|
3852 * window for this buffer in which 'spell' is set. */
|
|
236
|
3853 if (*buf->b_p_spl != NUL)
|
|
316
|
3854 {
|
|
|
3855 FOR_ALL_WINDOWS(wp)
|
|
|
3856 if (wp->w_buffer == buf && wp->w_p_spell)
|
|
|
3857 {
|
|
|
3858 (void)did_set_spelllang(buf);
|
|
|
3859 # ifdef FEAT_WINDOWS
|
|
|
3860 break;
|
|
|
3861 # endif
|
|
|
3862 }
|
|
|
3863 }
|
|
236
|
3864 }
|
|
|
3865 }
|
|
|
3866 # endif
|
|
|
3867
|
|
310
|
3868 /*
|
|
|
3869 * Reload the spell file "fname" if it's loaded.
|
|
|
3870 */
|
|
|
3871 static void
|
|
323
|
3872 spell_reload_one(fname, added_word)
|
|
310
|
3873 char_u *fname;
|
|
323
|
3874 int added_word; /* invoked through "zg" */
|
|
310
|
3875 {
|
|
|
3876 slang_T *lp;
|
|
323
|
3877 int didit = FALSE;
|
|
310
|
3878
|
|
|
3879 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
3880 if (fullpathcmp(fname, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
3881 {
|
|
|
3882 slang_clear(lp);
|
|
323
|
3883 (void)spell_load_file(fname, NULL, lp, FALSE);
|
|
310
|
3884 redraw_all_later(NOT_VALID);
|
|
323
|
3885 didit = TRUE;
|
|
310
|
3886 }
|
|
323
|
3887
|
|
|
3888 /* When "zg" was used and the file wasn't loaded yet, should redo
|
|
|
3889 * 'spelllang' to get it loaded. */
|
|
|
3890 if (added_word && !didit)
|
|
|
3891 did_set_spelllang(curbuf);
|
|
310
|
3892 }
|
|
|
3893
|
|
|
3894
|
|
236
|
3895 /*
|
|
|
3896 * Functions for ":mkspell".
|
|
|
3897 */
|
|
|
3898
|
|
300
|
3899 #define MAXLINELEN 500 /* Maximum length in bytes of a line in a .aff
|
|
236
|
3900 and .dic file. */
|
|
|
3901 /*
|
|
|
3902 * Main structure to store the contents of a ".aff" file.
|
|
|
3903 */
|
|
|
3904 typedef struct afffile_S
|
|
|
3905 {
|
|
|
3906 char_u *af_enc; /* "SET", normalized, alloc'ed string or NULL */
|
|
481
|
3907 int af_slash; /* character used in word for slash */
|
|
310
|
3908 int af_rar; /* RAR ID for rare word */
|
|
|
3909 int af_kep; /* KEP ID for keep-case word */
|
|
346
|
3910 int af_bad; /* BAD ID for banned word */
|
|
483
|
3911 int af_needaffix; /* NEEDAFFIX ID */
|
|
|
3912 char_u *af_compflags; /* COMPOUNDFLAG and COMPOUNDFLAGS concat'ed */
|
|
|
3913 int af_compmax; /* COMPOUNDMAX */
|
|
481
|
3914 int af_compminlen; /* COMPOUNDMIN */
|
|
483
|
3915 int af_compsylmax; /* COMPOUNDSYLMAX */
|
|
|
3916 char_u *af_syllable; /* SYLLABLE */
|
|
339
|
3917 int af_pfxpostpone; /* postpone prefixes without chop string */
|
|
236
|
3918 hashtab_T af_pref; /* hashtable for prefixes, affheader_T */
|
|
|
3919 hashtab_T af_suff; /* hashtable for suffixes, affheader_T */
|
|
|
3920 } afffile_T;
|
|
|
3921
|
|
|
3922 typedef struct affentry_S affentry_T;
|
|
|
3923 /* Affix entry from ".aff" file. Used for prefixes and suffixes. */
|
|
|
3924 struct affentry_S
|
|
|
3925 {
|
|
|
3926 affentry_T *ae_next; /* next affix with same name/number */
|
|
|
3927 char_u *ae_chop; /* text to chop off basic word (can be NULL) */
|
|
|
3928 char_u *ae_add; /* text to add to basic word (can be NULL) */
|
|
|
3929 char_u *ae_cond; /* condition (NULL for ".") */
|
|
|
3930 regprog_T *ae_prog; /* regexp program for ae_cond or NULL */
|
|
483
|
3931 char_u ae_rare; /* rare affix */
|
|
|
3932 char_u ae_nocomp; /* word with affix not compoundable */
|
|
300
|
3933 };
|
|
|
3934
|
|
455
|
3935 #define AH_KEY_LEN 10
|
|
|
3936
|
|
300
|
3937 /* Affix header from ".aff" file. Used for af_pref and af_suff. */
|
|
|
3938 typedef struct affheader_S
|
|
|
3939 {
|
|
455
|
3940 /* key for hashtable == name of affix entry */
|
|
|
3941 #ifdef FEAT_MBYTE
|
|
|
3942 char_u ah_key[AH_KEY_LEN]; /* multi-byte char plus NUL */
|
|
|
3943 #else
|
|
|
3944 char_u ah_key[2]; /* one byte char plus NUL */
|
|
|
3945 #endif
|
|
339
|
3946 int ah_newID; /* prefix ID after renumbering */
|
|
300
|
3947 int ah_combine; /* suffix may combine with prefix */
|
|
|
3948 affentry_T *ah_first; /* first affix entry */
|
|
|
3949 } affheader_T;
|
|
|
3950
|
|
|
3951 #define HI2AH(hi) ((affheader_T *)(hi)->hi_key)
|
|
|
3952
|
|
|
3953 /*
|
|
|
3954 * Structure that is used to store the items in the word tree. This avoids
|
|
474
|
3955 * the need to keep track of each allocated thing, everything is freed all at
|
|
|
3956 * once after ":mkspell" is done.
|
|
300
|
3957 */
|
|
|
3958 #define SBLOCKSIZE 16000 /* size of sb_data */
|
|
|
3959 typedef struct sblock_S sblock_T;
|
|
|
3960 struct sblock_S
|
|
|
3961 {
|
|
|
3962 sblock_T *sb_next; /* next block in list */
|
|
|
3963 int sb_used; /* nr of bytes already in use */
|
|
|
3964 char_u sb_data[1]; /* data, actually longer */
|
|
236
|
3965 };
|
|
|
3966
|
|
|
3967 /*
|
|
300
|
3968 * A node in the tree.
|
|
236
|
3969 */
|
|
300
|
3970 typedef struct wordnode_S wordnode_T;
|
|
|
3971 struct wordnode_S
|
|
236
|
3972 {
|
|
346
|
3973 union /* shared to save space */
|
|
|
3974 {
|
|
469
|
3975 char_u hashkey[6]; /* the hash key, only used while compressing */
|
|
346
|
3976 int index; /* index in written nodes (valid after first
|
|
|
3977 round) */
|
|
|
3978 } wn_u1;
|
|
|
3979 union /* shared to save space */
|
|
|
3980 {
|
|
|
3981 wordnode_T *next; /* next node with same hash key */
|
|
|
3982 wordnode_T *wnode; /* parent node that will write this node */
|
|
|
3983 } wn_u2;
|
|
300
|
3984 wordnode_T *wn_child; /* child (next byte in word) */
|
|
|
3985 wordnode_T *wn_sibling; /* next sibling (alternate byte in word,
|
|
|
3986 always sorted) */
|
|
474
|
3987 int wn_refs; /* Nr. of references to this node. Only
|
|
|
3988 relevant for first node in a list of
|
|
|
3989 siblings, in following siblings it is
|
|
|
3990 always one. */
|
|
300
|
3991 char_u wn_byte; /* Byte for this node. NUL for word end */
|
|
481
|
3992 char_u wn_affixID; /* when "wn_byte" is NUL: supported/required
|
|
469
|
3993 prefix ID or 0 */
|
|
|
3994 short_u wn_flags; /* when "wn_byte" is NUL: WF_ flags */
|
|
|
3995 short wn_region; /* when "wn_byte" is NUL: region mask; for
|
|
339
|
3996 PREFIXTREE it's the prefcondnr */
|
|
469
|
3997 #ifdef SPELL_PRINTTREE
|
|
|
3998 int wn_nr; /* sequence nr for printing */
|
|
|
3999 #endif
|
|
236
|
4000 };
|
|
|
4001
|
|
390
|
4002 #define WN_MASK 0xffff /* mask relevant bits of "wn_flags" */
|
|
|
4003
|
|
300
|
4004 #define HI2WN(hi) (wordnode_T *)((hi)->hi_key)
|
|
236
|
4005
|
|
300
|
4006 /*
|
|
|
4007 * Info used while reading the spell files.
|
|
|
4008 */
|
|
|
4009 typedef struct spellinfo_S
|
|
249
|
4010 {
|
|
300
|
4011 wordnode_T *si_foldroot; /* tree with case-folded words */
|
|
334
|
4012 long si_foldwcount; /* nr of words in si_foldroot */
|
|
474
|
4013
|
|
300
|
4014 wordnode_T *si_keeproot; /* tree with keep-case words */
|
|
334
|
4015 long si_keepwcount; /* nr of words in si_keeproot */
|
|
474
|
4016
|
|
339
|
4017 wordnode_T *si_prefroot; /* tree with postponed prefixes */
|
|
474
|
4018
|
|
300
|
4019 sblock_T *si_blocks; /* memory blocks used */
|
|
482
|
4020 long si_blocks_cnt; /* memory blocks allocated */
|
|
|
4021 long si_compress_cnt; /* words to add before lowering
|
|
|
4022 compression limit */
|
|
474
|
4023 wordnode_T *si_first_free; /* List of nodes that have been freed during
|
|
|
4024 compression, linked by "wn_child" field. */
|
|
482
|
4025 long si_free_count; /* number of nodes in si_first_free */
|
|
474
|
4026 #ifdef SPELL_PRINTTREE
|
|
|
4027 int si_wordnode_nr; /* sequence nr for nodes */
|
|
|
4028 #endif
|
|
|
4029
|
|
|
4030
|
|
300
|
4031 int si_ascii; /* handling only ASCII words */
|
|
310
|
4032 int si_add; /* addition file */
|
|
351
|
4033 int si_clear_chartab; /* when TRUE clear char tables */
|
|
300
|
4034 int si_region; /* region mask */
|
|
|
4035 vimconv_T si_conv; /* for conversion to 'encoding' */
|
|
302
|
4036 int si_memtot; /* runtime memory used */
|
|
310
|
4037 int si_verbose; /* verbose messages */
|
|
474
|
4038 int si_msg_count; /* number of words added since last message */
|
|
316
|
4039 int si_region_count; /* number of regions supported (1 when there
|
|
|
4040 are no regions) */
|
|
483
|
4041 char_u si_region_name[16]; /* region names; used only if
|
|
|
4042 * si_region_count > 1) */
|
|
323
|
4043
|
|
|
4044 garray_T si_rep; /* list of fromto_T entries from REP lines */
|
|
|
4045 garray_T si_sal; /* list of fromto_T entries from SAL lines */
|
|
375
|
4046 char_u *si_sofofr; /* SOFOFROM text */
|
|
|
4047 char_u *si_sofoto; /* SOFOTO text */
|
|
323
|
4048 int si_followup; /* soundsalike: ? */
|
|
|
4049 int si_collapse; /* soundsalike: ? */
|
|
|
4050 int si_rem_accents; /* soundsalike: remove accents */
|
|
|
4051 garray_T si_map; /* MAP info concatenated */
|
|
366
|
4052 char_u *si_midword; /* MIDWORD chars, alloc'ed string or NULL */
|
|
483
|
4053 int si_compmax; /* max nr of words for compounding */
|
|
481
|
4054 int si_compminlen; /* minimal length for compounding */
|
|
483
|
4055 int si_compsylmax; /* max nr of syllables for compounding */
|
|
481
|
4056 char_u *si_compflags; /* flags used for compounding */
|
|
492
|
4057 char_u si_nobreak; /* NOBREAK */
|
|
483
|
4058 char_u *si_syllable; /* syllable string */
|
|
339
|
4059 garray_T si_prefcond; /* table with conditions for postponed
|
|
|
4060 * prefixes, each stored as a string */
|
|
|
4061 int si_newID; /* current value for ah_newID */
|
|
300
|
4062 } spellinfo_T;
|
|
249
|
4063
|
|
474
|
4064 static afffile_T *spell_read_aff __ARGS((spellinfo_T *spin, char_u *fname));
|
|
339
|
4065 static int str_equal __ARGS((char_u *s1, char_u *s2));
|
|
323
|
4066 static void add_fromto __ARGS((spellinfo_T *spin, garray_T *gap, char_u *from, char_u *to));
|
|
|
4067 static int sal_to_bool __ARGS((char_u *s));
|
|
240
|
4068 static int has_non_ascii __ARGS((char_u *s));
|
|
300
|
4069 static void spell_free_aff __ARGS((afffile_T *aff));
|
|
474
|
4070 static int spell_read_dic __ARGS((spellinfo_T *spin, char_u *fname, afffile_T *affile));
|
|
483
|
4071 static int get_pfxlist __ARGS((afffile_T *affile, char_u *afflist, char_u *store_afflist));
|
|
|
4072 static void get_compflags __ARGS((spellinfo_T *spin, char_u *afflist, char_u *store_afflist));
|
|
|
4073 static int store_aff_word __ARGS((spellinfo_T *spin, char_u *word, char_u *afflist, afffile_T *affile, hashtab_T *ht, hashtab_T *xht, int comb, int flags, char_u *pfxlist, int pfxlen));
|
|
474
|
4074 static int spell_read_wordfile __ARGS((spellinfo_T *spin, char_u *fname));
|
|
|
4075 static void *getroom __ARGS((spellinfo_T *spin, size_t len, int align));
|
|
|
4076 static char_u *getroom_save __ARGS((spellinfo_T *spin, char_u *s));
|
|
300
|
4077 static void free_blocks __ARGS((sblock_T *bl));
|
|
474
|
4078 static wordnode_T *wordtree_alloc __ARGS((spellinfo_T *spin));
|
|
483
|
4079 static int store_word __ARGS((spellinfo_T *spin, char_u *word, int flags, int region, char_u *pfxlist, int need_affix));
|
|
481
|
4080 static int tree_add_word __ARGS((spellinfo_T *spin, char_u *word, wordnode_T *tree, int flags, int region, int affixID));
|
|
474
|
4081 static wordnode_T *get_wordnode __ARGS((spellinfo_T *spin));
|
|
|
4082 static void deref_wordnode __ARGS((spellinfo_T *spin, wordnode_T *node));
|
|
|
4083 static void free_wordnode __ARGS((spellinfo_T *spin, wordnode_T *n));
|
|
|
4084 static void wordtree_compress __ARGS((spellinfo_T *spin, wordnode_T *root));
|
|
|
4085 static int node_compress __ARGS((spellinfo_T *spin, wordnode_T *node, hashtab_T *ht, int *tot));
|
|
300
|
4086 static int node_equal __ARGS((wordnode_T *n1, wordnode_T *n2));
|
|
474
|
4087 static void write_vim_spell __ARGS((spellinfo_T *spin, char_u *fname));
|
|
346
|
4088 static void clear_node __ARGS((wordnode_T *node));
|
|
|
4089 static int put_node __ARGS((FILE *fd, wordnode_T *node, int index, int regionmask, int prefixtree));
|
|
323
|
4090 static void mkspell __ARGS((int fcount, char_u **fnames, int ascii, int overwrite, int added_word));
|
|
310
|
4091 static void init_spellfile __ARGS((void));
|
|
236
|
4092
|
|
455
|
4093 /* In the postponed prefixes tree wn_flags is used to store the WFP_ flags,
|
|
|
4094 * but it must be negative to indicate the prefix tree to tree_add_word().
|
|
|
4095 * Use a negative number with the lower 8 bits zero. */
|
|
|
4096 #define PFX_FLAGS -256
|
|
390
|
4097
|
|
483
|
4098 /*
|
|
|
4099 * Tunable parameters for when the tree is compressed. See 'mkspellmem'.
|
|
|
4100 */
|
|
|
4101 static long compress_start = 30000; /* memory / SBLOCKSIZE */
|
|
|
4102 static long compress_inc = 100; /* memory / SBLOCKSIZE */
|
|
|
4103 static long compress_added = 500000; /* word count */
|
|
|
4104
|
|
469
|
4105 #ifdef SPELL_PRINTTREE
|
|
|
4106 /*
|
|
|
4107 * For debugging the tree code: print the current tree in a (more or less)
|
|
|
4108 * readable format, so that we can see what happens when adding a word and/or
|
|
|
4109 * compressing the tree.
|
|
|
4110 * Based on code from Olaf Seibert.
|
|
|
4111 */
|
|
|
4112 #define PRINTLINESIZE 1000
|
|
|
4113 #define PRINTWIDTH 6
|
|
|
4114
|
|
|
4115 #define PRINTSOME(l, depth, fmt, a1, a2) vim_snprintf(l + depth * PRINTWIDTH, \
|
|
|
4116 PRINTLINESIZE - PRINTWIDTH * depth, fmt, a1, a2)
|
|
|
4117
|
|
|
4118 static char line1[PRINTLINESIZE];
|
|
|
4119 static char line2[PRINTLINESIZE];
|
|
|
4120 static char line3[PRINTLINESIZE];
|
|
|
4121
|
|
|
4122 static void
|
|
|
4123 spell_clear_flags(wordnode_T *node)
|
|
|
4124 {
|
|
|
4125 wordnode_T *np;
|
|
|
4126
|
|
|
4127 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
4128 {
|
|
|
4129 np->wn_u1.index = FALSE;
|
|
|
4130 spell_clear_flags(np->wn_child);
|
|
|
4131 }
|
|
|
4132 }
|
|
|
4133
|
|
|
4134 static void
|
|
|
4135 spell_print_node(wordnode_T *node, int depth)
|
|
|
4136 {
|
|
|
4137 if (node->wn_u1.index)
|
|
|
4138 {
|
|
|
4139 /* Done this node before, print the reference. */
|
|
|
4140 PRINTSOME(line1, depth, "(%d)", node->wn_nr, 0);
|
|
|
4141 PRINTSOME(line2, depth, " ", 0, 0);
|
|
|
4142 PRINTSOME(line3, depth, " ", 0, 0);
|
|
|
4143 msg(line1);
|
|
|
4144 msg(line2);
|
|
|
4145 msg(line3);
|
|
|
4146 }
|
|
|
4147 else
|
|
|
4148 {
|
|
|
4149 node->wn_u1.index = TRUE;
|
|
|
4150
|
|
|
4151 if (node->wn_byte != NUL)
|
|
|
4152 {
|
|
|
4153 if (node->wn_child != NULL)
|
|
|
4154 PRINTSOME(line1, depth, " %c -> ", node->wn_byte, 0);
|
|
|
4155 else
|
|
|
4156 /* Cannot happen? */
|
|
|
4157 PRINTSOME(line1, depth, " %c ???", node->wn_byte, 0);
|
|
|
4158 }
|
|
|
4159 else
|
|
|
4160 PRINTSOME(line1, depth, " $ ", 0, 0);
|
|
|
4161
|
|
|
4162 PRINTSOME(line2, depth, "%d/%d ", node->wn_nr, node->wn_refs);
|
|
|
4163
|
|
|
4164 if (node->wn_sibling != NULL)
|
|
|
4165 PRINTSOME(line3, depth, " | ", 0, 0);
|
|
|
4166 else
|
|
|
4167 PRINTSOME(line3, depth, " ", 0, 0);
|
|
|
4168
|
|
|
4169 if (node->wn_byte == NUL)
|
|
|
4170 {
|
|
|
4171 msg(line1);
|
|
|
4172 msg(line2);
|
|
|
4173 msg(line3);
|
|
|
4174 }
|
|
|
4175
|
|
|
4176 /* do the children */
|
|
|
4177 if (node->wn_byte != NUL && node->wn_child != NULL)
|
|
|
4178 spell_print_node(node->wn_child, depth + 1);
|
|
|
4179
|
|
|
4180 /* do the siblings */
|
|
|
4181 if (node->wn_sibling != NULL)
|
|
|
4182 {
|
|
|
4183 /* get rid of all parent details except | */
|
|
|
4184 STRCPY(line1, line3);
|
|
|
4185 STRCPY(line2, line3);
|
|
|
4186 spell_print_node(node->wn_sibling, depth);
|
|
|
4187 }
|
|
|
4188 }
|
|
|
4189 }
|
|
|
4190
|
|
|
4191 static void
|
|
|
4192 spell_print_tree(wordnode_T *root)
|
|
|
4193 {
|
|
|
4194 if (root != NULL)
|
|
|
4195 {
|
|
|
4196 /* Clear the "wn_u1.index" fields, used to remember what has been
|
|
|
4197 * done. */
|
|
|
4198 spell_clear_flags(root);
|
|
|
4199
|
|
|
4200 /* Recursively print the tree. */
|
|
|
4201 spell_print_node(root, 0);
|
|
|
4202 }
|
|
|
4203 }
|
|
|
4204 #endif /* SPELL_PRINTTREE */
|
|
|
4205
|
|
236
|
4206 /*
|
|
323
|
4207 * Read the affix file "fname".
|
|
316
|
4208 * Returns an afffile_T, NULL for complete failure.
|
|
236
|
4209 */
|
|
|
4210 static afffile_T *
|
|
474
|
4211 spell_read_aff(spin, fname)
|
|
|
4212 spellinfo_T *spin;
|
|
236
|
4213 char_u *fname;
|
|
|
4214 {
|
|
|
4215 FILE *fd;
|
|
|
4216 afffile_T *aff;
|
|
|
4217 char_u rline[MAXLINELEN];
|
|
|
4218 char_u *line;
|
|
|
4219 char_u *pc = NULL;
|
|
334
|
4220 #define MAXITEMCNT 7
|
|
|
4221 char_u *(items[MAXITEMCNT]);
|
|
236
|
4222 int itemcnt;
|
|
|
4223 char_u *p;
|
|
|
4224 int lnum = 0;
|
|
|
4225 affheader_T *cur_aff = NULL;
|
|
|
4226 int aff_todo = 0;
|
|
|
4227 hashtab_T *tp;
|
|
255
|
4228 char_u *low = NULL;
|
|
|
4229 char_u *fol = NULL;
|
|
|
4230 char_u *upp = NULL;
|
|
307
|
4231 static char *e_affname = N_("Affix name too long in %s line %d: %s");
|
|
323
|
4232 int do_rep;
|
|
|
4233 int do_sal;
|
|
|
4234 int do_map;
|
|
366
|
4235 int do_midword;
|
|
375
|
4236 int do_sofo;
|
|
323
|
4237 int found_map = FALSE;
|
|
324
|
4238 hashitem_T *hi;
|
|
455
|
4239 int l;
|
|
236
|
4240
|
|
300
|
4241 /*
|
|
|
4242 * Open the file.
|
|
|
4243 */
|
|
310
|
4244 fd = mch_fopen((char *)fname, "r");
|
|
236
|
4245 if (fd == NULL)
|
|
|
4246 {
|
|
|
4247 EMSG2(_(e_notopen), fname);
|
|
|
4248 return NULL;
|
|
|
4249 }
|
|
|
4250
|
|
310
|
4251 if (spin->si_verbose || p_verbose > 2)
|
|
|
4252 {
|
|
|
4253 if (!spin->si_verbose)
|
|
|
4254 verbose_enter();
|
|
366
|
4255 smsg((char_u *)_("Reading affix file %s ..."), fname);
|
|
310
|
4256 out_flush();
|
|
|
4257 if (!spin->si_verbose)
|
|
|
4258 verbose_leave();
|
|
|
4259 }
|
|
236
|
4260
|
|
323
|
4261 /* Only do REP lines when not done in another .aff file already. */
|
|
|
4262 do_rep = spin->si_rep.ga_len == 0;
|
|
|
4263
|
|
|
4264 /* Only do SAL lines when not done in another .aff file already. */
|
|
|
4265 do_sal = spin->si_sal.ga_len == 0;
|
|
|
4266
|
|
|
4267 /* Only do MAP lines when not done in another .aff file already. */
|
|
|
4268 do_map = spin->si_map.ga_len == 0;
|
|
|
4269
|
|
366
|
4270 /* Only do MIDWORD line when not done in another .aff file already */
|
|
|
4271 do_midword = spin->si_midword == NULL;
|
|
|
4272
|
|
375
|
4273 /* Only do SOFOFROM and SOFOTO when not done in another .aff file already */
|
|
|
4274 do_sofo = spin->si_sofofr == NULL;
|
|
|
4275
|
|
300
|
4276 /*
|
|
|
4277 * Allocate and init the afffile_T structure.
|
|
|
4278 */
|
|
474
|
4279 aff = (afffile_T *)getroom(spin, sizeof(afffile_T), TRUE);
|
|
236
|
4280 if (aff == NULL)
|
|
|
4281 return NULL;
|
|
|
4282 hash_init(&aff->af_pref);
|
|
|
4283 hash_init(&aff->af_suff);
|
|
|
4284
|
|
|
4285 /*
|
|
|
4286 * Read all the lines in the file one by one.
|
|
|
4287 */
|
|
255
|
4288 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int)
|
|
236
|
4289 {
|
|
255
|
4290 line_breakcheck();
|
|
236
|
4291 ++lnum;
|
|
|
4292
|
|
|
4293 /* Skip comment lines. */
|
|
|
4294 if (*rline == '#')
|
|
|
4295 continue;
|
|
|
4296
|
|
|
4297 /* Convert from "SET" to 'encoding' when needed. */
|
|
|
4298 vim_free(pc);
|
|
310
|
4299 #ifdef FEAT_MBYTE
|
|
300
|
4300 if (spin->si_conv.vc_type != CONV_NONE)
|
|
236
|
4301 {
|
|
300
|
4302 pc = string_convert(&spin->si_conv, rline, NULL);
|
|
255
|
4303 if (pc == NULL)
|
|
|
4304 {
|
|
|
4305 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
4306 fname, lnum, rline);
|
|
|
4307 continue;
|
|
|
4308 }
|
|
236
|
4309 line = pc;
|
|
|
4310 }
|
|
|
4311 else
|
|
310
|
4312 #endif
|
|
236
|
4313 {
|
|
|
4314 pc = NULL;
|
|
|
4315 line = rline;
|
|
|
4316 }
|
|
|
4317
|
|
|
4318 /* Split the line up in white separated items. Put a NUL after each
|
|
|
4319 * item. */
|
|
|
4320 itemcnt = 0;
|
|
|
4321 for (p = line; ; )
|
|
|
4322 {
|
|
|
4323 while (*p != NUL && *p <= ' ') /* skip white space and CR/NL */
|
|
|
4324 ++p;
|
|
|
4325 if (*p == NUL)
|
|
|
4326 break;
|
|
334
|
4327 if (itemcnt == MAXITEMCNT) /* too many items */
|
|
300
|
4328 break;
|
|
236
|
4329 items[itemcnt++] = p;
|
|
300
|
4330 while (*p > ' ') /* skip until white space or CR/NL */
|
|
236
|
4331 ++p;
|
|
|
4332 if (*p == NUL)
|
|
|
4333 break;
|
|
|
4334 *p++ = NUL;
|
|
|
4335 }
|
|
|
4336
|
|
|
4337 /* Handle non-empty lines. */
|
|
|
4338 if (itemcnt > 0)
|
|
|
4339 {
|
|
|
4340 if (STRCMP(items[0], "SET") == 0 && itemcnt == 2
|
|
|
4341 && aff->af_enc == NULL)
|
|
|
4342 {
|
|
310
|
4343 #ifdef FEAT_MBYTE
|
|
300
|
4344 /* Setup for conversion from "ENC" to 'encoding'. */
|
|
|
4345 aff->af_enc = enc_canonize(items[1]);
|
|
|
4346 if (aff->af_enc != NULL && !spin->si_ascii
|
|
|
4347 && convert_setup(&spin->si_conv, aff->af_enc,
|
|
|
4348 p_enc) == FAIL)
|
|
|
4349 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"),
|
|
|
4350 fname, aff->af_enc, p_enc);
|
|
375
|
4351 spin->si_conv.vc_fail = TRUE;
|
|
310
|
4352 #else
|
|
|
4353 smsg((char_u *)_("Conversion in %s not supported"), fname);
|
|
|
4354 #endif
|
|
236
|
4355 }
|
|
366
|
4356 else if (STRCMP(items[0], "MIDWORD") == 0 && itemcnt == 2)
|
|
|
4357 {
|
|
|
4358 if (do_midword)
|
|
|
4359 spin->si_midword = vim_strsave(items[1]);
|
|
|
4360 }
|
|
302
|
4361 else if (STRCMP(items[0], "NOSPLITSUGS") == 0 && itemcnt == 1)
|
|
|
4362 {
|
|
323
|
4363 /* ignored, we always split */
|
|
302
|
4364 }
|
|
323
|
4365 else if (STRCMP(items[0], "TRY") == 0 && itemcnt == 2)
|
|
300
|
4366 {
|
|
323
|
4367 /* ignored, we look in the tree for what chars may appear */
|
|
300
|
4368 }
|
|
481
|
4369 else if (STRCMP(items[0], "SLASH") == 0 && itemcnt == 2
|
|
|
4370 && aff->af_slash == 0)
|
|
|
4371 {
|
|
|
4372 aff->af_slash = items[1][0];
|
|
|
4373 if (items[1][1] != NUL)
|
|
|
4374 smsg((char_u *)_("Character used for SLASH must be ASCII; in %s line %d: %s"),
|
|
|
4375 fname, lnum, items[1]);
|
|
|
4376 }
|
|
307
|
4377 else if (STRCMP(items[0], "RAR") == 0 && itemcnt == 2
|
|
|
4378 && aff->af_rar == 0)
|
|
|
4379 {
|
|
|
4380 aff->af_rar = items[1][0];
|
|
|
4381 if (items[1][1] != NUL)
|
|
|
4382 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
4383 }
|
|
310
|
4384 else if (STRCMP(items[0], "KEP") == 0 && itemcnt == 2
|
|
|
4385 && aff->af_kep == 0)
|
|
307
|
4386 {
|
|
310
|
4387 aff->af_kep = items[1][0];
|
|
307
|
4388 if (items[1][1] != NUL)
|
|
|
4389 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
4390 }
|
|
346
|
4391 else if (STRCMP(items[0], "BAD") == 0 && itemcnt == 2
|
|
|
4392 && aff->af_bad == 0)
|
|
|
4393 {
|
|
|
4394 aff->af_bad = items[1][0];
|
|
|
4395 if (items[1][1] != NUL)
|
|
|
4396 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
4397 }
|
|
483
|
4398 else if (STRCMP(items[0], "NEEDAFFIX") == 0 && itemcnt == 2
|
|
|
4399 && aff->af_needaffix == 0)
|
|
|
4400 {
|
|
|
4401 aff->af_needaffix = items[1][0];
|
|
|
4402 if (items[1][1] != NUL)
|
|
|
4403 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
4404 }
|
|
481
|
4405 else if (STRCMP(items[0], "COMPOUNDFLAG") == 0 && itemcnt == 2
|
|
483
|
4406 && aff->af_compflags == NULL)
|
|
|
4407 {
|
|
|
4408 p = getroom(spin, 3, FALSE);
|
|
|
4409 if (p != NULL)
|
|
|
4410 {
|
|
|
4411 /* Turn single flag "c" into COMPOUNDFLAGS compatible
|
|
|
4412 * string "c+". */
|
|
|
4413 p[0] = items[1][0];
|
|
|
4414 p[1] = '+';
|
|
|
4415 p[2] = NUL;
|
|
|
4416 aff->af_compflags = p;
|
|
|
4417 }
|
|
481
|
4418 if (items[1][1] != NUL)
|
|
|
4419 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
4420 }
|
|
483
|
4421 else if (STRCMP(items[0], "COMPOUNDFLAGS") == 0 && itemcnt == 2)
|
|
|
4422 {
|
|
|
4423 /* Concatenate this string to previously defined ones, using a
|
|
|
4424 * slash to separate them. */
|
|
|
4425 l = STRLEN(items[1]) + 1;
|
|
|
4426 if (aff->af_compflags != NULL)
|
|
|
4427 l += STRLEN(aff->af_compflags) + 1;
|
|
|
4428 p = getroom(spin, l, FALSE);
|
|
|
4429 if (p != NULL)
|
|
|
4430 {
|
|
|
4431 if (aff->af_compflags != NULL)
|
|
|
4432 {
|
|
|
4433 STRCPY(p, aff->af_compflags);
|
|
|
4434 STRCAT(p, "/");
|
|
|
4435 }
|
|
|
4436 STRCAT(p, items[1]);
|
|
|
4437 aff->af_compflags = p;
|
|
|
4438 }
|
|
|
4439 }
|
|
|
4440 else if (STRCMP(items[0], "COMPOUNDMAX") == 0 && itemcnt == 2
|
|
|
4441 && aff->af_compmax == 0)
|
|
|
4442 {
|
|
|
4443 aff->af_compmax = atoi((char *)items[1]);
|
|
|
4444 if (aff->af_compmax == 0)
|
|
|
4445 smsg((char_u *)_("Wrong COMPOUNDMAX value in %s line %d: %s"),
|
|
|
4446 fname, lnum, items[1]);
|
|
481
|
4447 }
|
|
|
4448 else if (STRCMP(items[0], "COMPOUNDMIN") == 0 && itemcnt == 2
|
|
|
4449 && aff->af_compminlen == 0)
|
|
|
4450 {
|
|
|
4451 aff->af_compminlen = atoi((char *)items[1]);
|
|
|
4452 if (aff->af_compminlen == 0)
|
|
|
4453 smsg((char_u *)_("Wrong COMPOUNDMIN value in %s line %d: %s"),
|
|
|
4454 fname, lnum, items[1]);
|
|
|
4455 }
|
|
483
|
4456 else if (STRCMP(items[0], "COMPOUNDSYLMAX") == 0 && itemcnt == 2
|
|
|
4457 && aff->af_compsylmax == 0)
|
|
|
4458 {
|
|
|
4459 aff->af_compsylmax = atoi((char *)items[1]);
|
|
|
4460 if (aff->af_compsylmax == 0)
|
|
|
4461 smsg((char_u *)_("Wrong COMPOUNDSYLMAX value in %s line %d: %s"),
|
|
|
4462 fname, lnum, items[1]);
|
|
|
4463 }
|
|
|
4464 else if (STRCMP(items[0], "SYLLABLE") == 0 && itemcnt == 2
|
|
|
4465 && aff->af_syllable == NULL)
|
|
|
4466 {
|
|
|
4467 aff->af_syllable = getroom_save(spin, items[1]);
|
|
|
4468 }
|
|
492
|
4469 else if (STRCMP(items[0], "NOBREAK") == 0 && itemcnt == 1)
|
|
|
4470 {
|
|
|
4471 spin->si_nobreak = TRUE;
|
|
|
4472 }
|
|
339
|
4473 else if (STRCMP(items[0], "PFXPOSTPONE") == 0 && itemcnt == 1)
|
|
|
4474 {
|
|
|
4475 aff->af_pfxpostpone = TRUE;
|
|
|
4476 }
|
|
236
|
4477 else if ((STRCMP(items[0], "PFX") == 0
|
|
|
4478 || STRCMP(items[0], "SFX") == 0)
|
|
|
4479 && aff_todo == 0
|
|
334
|
4480 && itemcnt >= 4)
|
|
236
|
4481 {
|
|
334
|
4482 /* Myspell allows extra text after the item, but that might
|
|
|
4483 * mean mistakes go unnoticed. Require a comment-starter. */
|
|
|
4484 if (itemcnt > 4 && *items[4] != '#')
|
|
|
4485 smsg((char_u *)_("Trailing text in %s line %d: %s"),
|
|
|
4486 fname, lnum, items[4]);
|
|
|
4487
|
|
236
|
4488 /* New affix letter. */
|
|
474
|
4489 cur_aff = (affheader_T *)getroom(spin,
|
|
447
|
4490 sizeof(affheader_T), TRUE);
|
|
236
|
4491 if (cur_aff == NULL)
|
|
|
4492 break;
|
|
455
|
4493 #ifdef FEAT_MBYTE
|
|
|
4494 if (has_mbyte)
|
|
|
4495 {
|
|
474
|
4496 l = (*mb_ptr2len)(items[1]);
|
|
455
|
4497 if (l >= AH_KEY_LEN)
|
|
|
4498 l = 1; /* too long, must be an overlong sequence */
|
|
|
4499 else
|
|
|
4500 mch_memmove(cur_aff->ah_key, items[1], l);
|
|
|
4501 }
|
|
|
4502 else
|
|
|
4503 #endif
|
|
|
4504 {
|
|
|
4505 *cur_aff->ah_key = *items[1];
|
|
|
4506 l = 1;
|
|
|
4507 }
|
|
|
4508 cur_aff->ah_key[l] = NUL;
|
|
|
4509 if (items[1][l] != NUL)
|
|
307
|
4510 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
236
|
4511 if (*items[2] == 'Y')
|
|
|
4512 cur_aff->ah_combine = TRUE;
|
|
300
|
4513 else if (*items[2] != 'N')
|
|
236
|
4514 smsg((char_u *)_("Expected Y or N in %s line %d: %s"),
|
|
|
4515 fname, lnum, items[2]);
|
|
339
|
4516
|
|
236
|
4517 if (*items[0] == 'P')
|
|
339
|
4518 {
|
|
236
|
4519 tp = &aff->af_pref;
|
|
339
|
4520 /* Use a new number in the .spl file later, to be able to
|
|
|
4521 * handle multiple .aff files. */
|
|
|
4522 if (aff->af_pfxpostpone)
|
|
344
|
4523 cur_aff->ah_newID = ++spin->si_newID;
|
|
339
|
4524 }
|
|
236
|
4525 else
|
|
|
4526 tp = &aff->af_suff;
|
|
300
|
4527 aff_todo = atoi((char *)items[3]);
|
|
324
|
4528 hi = hash_find(tp, cur_aff->ah_key);
|
|
|
4529 if (!HASHITEM_EMPTY(hi))
|
|
300
|
4530 {
|
|
236
|
4531 smsg((char_u *)_("Duplicate affix in %s line %d: %s"),
|
|
|
4532 fname, lnum, items[1]);
|
|
300
|
4533 aff_todo = 0;
|
|
|
4534 }
|
|
236
|
4535 else
|
|
|
4536 hash_add(tp, cur_aff->ah_key);
|
|
|
4537 }
|
|
|
4538 else if ((STRCMP(items[0], "PFX") == 0
|
|
|
4539 || STRCMP(items[0], "SFX") == 0)
|
|
|
4540 && aff_todo > 0
|
|
|
4541 && STRCMP(cur_aff->ah_key, items[1]) == 0
|
|
334
|
4542 && itemcnt >= 5)
|
|
236
|
4543 {
|
|
|
4544 affentry_T *aff_entry;
|
|
366
|
4545 int rare = FALSE;
|
|
483
|
4546 int nocomp = FALSE;
|
|
455
|
4547 int upper = FALSE;
|
|
366
|
4548 int lasti = 5;
|
|
|
4549
|
|
483
|
4550 /* Check for "rare" and "nocomp" after the other info. */
|
|
|
4551 while (itemcnt > lasti)
|
|
|
4552 {
|
|
|
4553 if (!rare && STRICMP(items[lasti], "rare") == 0)
|
|
|
4554 {
|
|
|
4555 rare = TRUE;
|
|
|
4556 ++lasti;
|
|
|
4557 }
|
|
|
4558 else if (!nocomp && STRICMP(items[lasti], "nocomp") == 0)
|
|
|
4559 {
|
|
|
4560 nocomp = TRUE;
|
|
|
4561 ++lasti;
|
|
|
4562 }
|
|
|
4563 else
|
|
|
4564 break;
|
|
366
|
4565 }
|
|
236
|
4566
|
|
334
|
4567 /* Myspell allows extra text after the item, but that might
|
|
|
4568 * mean mistakes go unnoticed. Require a comment-starter. */
|
|
366
|
4569 if (itemcnt > lasti && *items[lasti] != '#')
|
|
482
|
4570 smsg((char_u *)_(e_afftrailing), fname, lnum, items[lasti]);
|
|
334
|
4571
|
|
236
|
4572 /* New item for an affix letter. */
|
|
|
4573 --aff_todo;
|
|
474
|
4574 aff_entry = (affentry_T *)getroom(spin,
|
|
447
|
4575 sizeof(affentry_T), TRUE);
|
|
236
|
4576 if (aff_entry == NULL)
|
|
|
4577 break;
|
|
366
|
4578 aff_entry->ae_rare = rare;
|
|
483
|
4579 aff_entry->ae_nocomp = nocomp;
|
|
240
|
4580
|
|
236
|
4581 if (STRCMP(items[2], "0") != 0)
|
|
474
|
4582 aff_entry->ae_chop = getroom_save(spin, items[2]);
|
|
236
|
4583 if (STRCMP(items[3], "0") != 0)
|
|
474
|
4584 aff_entry->ae_add = getroom_save(spin, items[3]);
|
|
236
|
4585
|
|
300
|
4586 /* Don't use an affix entry with non-ASCII characters when
|
|
|
4587 * "spin->si_ascii" is TRUE. */
|
|
|
4588 if (!spin->si_ascii || !(has_non_ascii(aff_entry->ae_chop)
|
|
240
|
4589 || has_non_ascii(aff_entry->ae_add)))
|
|
|
4590 {
|
|
|
4591 aff_entry->ae_next = cur_aff->ah_first;
|
|
|
4592 cur_aff->ah_first = aff_entry;
|
|
300
|
4593
|
|
|
4594 if (STRCMP(items[4], ".") != 0)
|
|
|
4595 {
|
|
|
4596 char_u buf[MAXLINELEN];
|
|
|
4597
|
|
474
|
4598 aff_entry->ae_cond = getroom_save(spin, items[4]);
|
|
300
|
4599 if (*items[0] == 'P')
|
|
|
4600 sprintf((char *)buf, "^%s", items[4]);
|
|
|
4601 else
|
|
|
4602 sprintf((char *)buf, "%s$", items[4]);
|
|
|
4603 aff_entry->ae_prog = vim_regcomp(buf,
|
|
481
|
4604 RE_MAGIC + RE_STRING + RE_STRICT);
|
|
|
4605 if (aff_entry->ae_prog == NULL)
|
|
|
4606 smsg((char_u *)_("Broken condition in %s line %d: %s"),
|
|
|
4607 fname, lnum, items[4]);
|
|
300
|
4608 }
|
|
339
|
4609
|
|
|
4610 /* For postponed prefixes we need an entry in si_prefcond
|
|
|
4611 * for the condition. Use an existing one if possible. */
|
|
455
|
4612 if (*items[0] == 'P' && aff->af_pfxpostpone)
|
|
339
|
4613 {
|
|
455
|
4614 /* When the chop string is one lower-case letter and
|
|
|
4615 * the add string ends in the upper-case letter we set
|
|
|
4616 * the "upper" flag, clear "ae_chop" and remove the
|
|
|
4617 * letters from "ae_add". The condition must either
|
|
|
4618 * be empty or start with the same letter. */
|
|
|
4619 if (aff_entry->ae_chop != NULL
|
|
|
4620 && aff_entry->ae_add != NULL
|
|
|
4621 #ifdef FEAT_MBYTE
|
|
474
|
4622 && aff_entry->ae_chop[(*mb_ptr2len)(
|
|
455
|
4623 aff_entry->ae_chop)] == NUL
|
|
|
4624 #else
|
|
|
4625 && aff_entry->ae_chop[1] == NUL
|
|
|
4626 #endif
|
|
|
4627 )
|
|
339
|
4628 {
|
|
455
|
4629 int c, c_up;
|
|
|
4630
|
|
|
4631 c = PTR2CHAR(aff_entry->ae_chop);
|
|
|
4632 c_up = SPELL_TOUPPER(c);
|
|
|
4633 if (c_up != c
|
|
|
4634 && (aff_entry->ae_cond == NULL
|
|
|
4635 || PTR2CHAR(aff_entry->ae_cond) == c))
|
|
|
4636 {
|
|
|
4637 p = aff_entry->ae_add
|
|
|
4638 + STRLEN(aff_entry->ae_add);
|
|
|
4639 mb_ptr_back(aff_entry->ae_add, p);
|
|
|
4640 if (PTR2CHAR(p) == c_up)
|
|
|
4641 {
|
|
|
4642 upper = TRUE;
|
|
|
4643 aff_entry->ae_chop = NULL;
|
|
|
4644 *p = NUL;
|
|
|
4645
|
|
|
4646 /* The condition is matched with the
|
|
|
4647 * actual word, thus must check for the
|
|
|
4648 * upper-case letter. */
|
|
|
4649 if (aff_entry->ae_cond != NULL)
|
|
|
4650 {
|
|
|
4651 char_u buf[MAXLINELEN];
|
|
|
4652 #ifdef FEAT_MBYTE
|
|
|
4653 if (has_mbyte)
|
|
|
4654 {
|
|
|
4655 onecap_copy(items[4], buf, TRUE);
|
|
|
4656 aff_entry->ae_cond = getroom_save(
|
|
474
|
4657 spin, buf);
|
|
455
|
4658 }
|
|
|
4659 else
|
|
|
4660 #endif
|
|
|
4661 *aff_entry->ae_cond = c_up;
|
|
|
4662 if (aff_entry->ae_cond != NULL)
|
|
|
4663 {
|
|
|
4664 sprintf((char *)buf, "^%s",
|
|
339
|
4665 aff_entry->ae_cond);
|
|
455
|
4666 vim_free(aff_entry->ae_prog);
|
|
|
4667 aff_entry->ae_prog = vim_regcomp(
|
|
|
4668 buf, RE_MAGIC + RE_STRING);
|
|
|
4669 }
|
|
|
4670 }
|
|
|
4671 }
|
|
|
4672 }
|
|
339
|
4673 }
|
|
|
4674
|
|
455
|
4675 if (aff_entry->ae_chop == NULL)
|
|
390
|
4676 {
|
|
455
|
4677 int idx;
|
|
|
4678 char_u **pp;
|
|
|
4679 int n;
|
|
|
4680
|
|
|
4681 for (idx = spin->si_prefcond.ga_len - 1; idx >= 0;
|
|
|
4682 --idx)
|
|
|
4683 {
|
|
|
4684 p = ((char_u **)spin->si_prefcond.ga_data)[idx];
|
|
|
4685 if (str_equal(p, aff_entry->ae_cond))
|
|
|
4686 break;
|
|
|
4687 }
|
|
|
4688 if (idx < 0 && ga_grow(&spin->si_prefcond, 1) == OK)
|
|
|
4689 {
|
|
|
4690 /* Not found, add a new condition. */
|
|
|
4691 idx = spin->si_prefcond.ga_len++;
|
|
|
4692 pp = ((char_u **)spin->si_prefcond.ga_data)
|
|
|
4693 + idx;
|
|
|
4694 if (aff_entry->ae_cond == NULL)
|
|
|
4695 *pp = NULL;
|
|
|
4696 else
|
|
474
|
4697 *pp = getroom_save(spin,
|
|
455
|
4698 aff_entry->ae_cond);
|
|
|
4699 }
|
|
|
4700
|
|
|
4701 /* Add the prefix to the prefix tree. */
|
|
|
4702 if (aff_entry->ae_add == NULL)
|
|
|
4703 p = (char_u *)"";
|
|
390
|
4704 else
|
|
455
|
4705 p = aff_entry->ae_add;
|
|
|
4706 /* PFX_FLAGS is a negative number, so that
|
|
|
4707 * tree_add_word() knows this is the prefix tree. */
|
|
|
4708 n = PFX_FLAGS;
|
|
|
4709 if (rare)
|
|
|
4710 n |= WFP_RARE;
|
|
|
4711 if (!cur_aff->ah_combine)
|
|
|
4712 n |= WFP_NC;
|
|
|
4713 if (upper)
|
|
|
4714 n |= WFP_UP;
|
|
474
|
4715 tree_add_word(spin, p, spin->si_prefroot, n,
|
|
|
4716 idx, cur_aff->ah_newID);
|
|
390
|
4717 }
|
|
339
|
4718 }
|
|
240
|
4719 }
|
|
236
|
4720 }
|
|
255
|
4721 else if (STRCMP(items[0], "FOL") == 0 && itemcnt == 2)
|
|
|
4722 {
|
|
|
4723 if (fol != NULL)
|
|
|
4724 smsg((char_u *)_("Duplicate FOL in %s line %d"),
|
|
|
4725 fname, lnum);
|
|
|
4726 else
|
|
|
4727 fol = vim_strsave(items[1]);
|
|
|
4728 }
|
|
|
4729 else if (STRCMP(items[0], "LOW") == 0 && itemcnt == 2)
|
|
|
4730 {
|
|
|
4731 if (low != NULL)
|
|
|
4732 smsg((char_u *)_("Duplicate LOW in %s line %d"),
|
|
|
4733 fname, lnum);
|
|
|
4734 else
|
|
|
4735 low = vim_strsave(items[1]);
|
|
|
4736 }
|
|
|
4737 else if (STRCMP(items[0], "UPP") == 0 && itemcnt == 2)
|
|
|
4738 {
|
|
|
4739 if (upp != NULL)
|
|
|
4740 smsg((char_u *)_("Duplicate UPP in %s line %d"),
|
|
|
4741 fname, lnum);
|
|
|
4742 else
|
|
|
4743 upp = vim_strsave(items[1]);
|
|
|
4744 }
|
|
236
|
4745 else if (STRCMP(items[0], "REP") == 0 && itemcnt == 2)
|
|
323
|
4746 {
|
|
236
|
4747 /* Ignore REP count */;
|
|
323
|
4748 if (!isdigit(*items[1]))
|
|
|
4749 smsg((char_u *)_("Expected REP count in %s line %d"),
|
|
|
4750 fname, lnum);
|
|
|
4751 }
|
|
482
|
4752 else if (STRCMP(items[0], "REP") == 0 && itemcnt >= 3)
|
|
236
|
4753 {
|
|
|
4754 /* REP item */
|
|
482
|
4755 /* Myspell ignores extra arguments, we require it starts with
|
|
|
4756 * # to detect mistakes. */
|
|
|
4757 if (itemcnt > 3 && items[3][0] != '#')
|
|
|
4758 smsg((char_u *)_(e_afftrailing), fname, lnum, items[3]);
|
|
323
|
4759 if (do_rep)
|
|
|
4760 add_fromto(spin, &spin->si_rep, items[1], items[2]);
|
|
|
4761 }
|
|
|
4762 else if (STRCMP(items[0], "MAP") == 0 && itemcnt == 2)
|
|
|
4763 {
|
|
|
4764 /* MAP item or count */
|
|
|
4765 if (!found_map)
|
|
|
4766 {
|
|
|
4767 /* First line contains the count. */
|
|
|
4768 found_map = TRUE;
|
|
|
4769 if (!isdigit(*items[1]))
|
|
|
4770 smsg((char_u *)_("Expected MAP count in %s line %d"),
|
|
|
4771 fname, lnum);
|
|
|
4772 }
|
|
|
4773 else if (do_map)
|
|
|
4774 {
|
|
346
|
4775 int c;
|
|
|
4776
|
|
|
4777 /* Check that every character appears only once. */
|
|
|
4778 for (p = items[1]; *p != NUL; )
|
|
|
4779 {
|
|
|
4780 #ifdef FEAT_MBYTE
|
|
|
4781 c = mb_ptr2char_adv(&p);
|
|
|
4782 #else
|
|
|
4783 c = *p++;
|
|
|
4784 #endif
|
|
|
4785 if ((spin->si_map.ga_len > 0
|
|
|
4786 && vim_strchr(spin->si_map.ga_data, c)
|
|
|
4787 != NULL)
|
|
|
4788 || vim_strchr(p, c) != NULL)
|
|
|
4789 smsg((char_u *)_("Duplicate character in MAP in %s line %d"),
|
|
|
4790 fname, lnum);
|
|
|
4791 }
|
|
|
4792
|
|
323
|
4793 /* We simply concatenate all the MAP strings, separated by
|
|
|
4794 * slashes. */
|
|
|
4795 ga_concat(&spin->si_map, items[1]);
|
|
|
4796 ga_append(&spin->si_map, '/');
|
|
|
4797 }
|
|
|
4798 }
|
|
|
4799 else if (STRCMP(items[0], "SAL") == 0 && itemcnt == 3)
|
|
|
4800 {
|
|
|
4801 if (do_sal)
|
|
|
4802 {
|
|
|
4803 /* SAL item (sounds-a-like)
|
|
|
4804 * Either one of the known keys or a from-to pair. */
|
|
|
4805 if (STRCMP(items[1], "followup") == 0)
|
|
|
4806 spin->si_followup = sal_to_bool(items[2]);
|
|
|
4807 else if (STRCMP(items[1], "collapse_result") == 0)
|
|
|
4808 spin->si_collapse = sal_to_bool(items[2]);
|
|
|
4809 else if (STRCMP(items[1], "remove_accents") == 0)
|
|
|
4810 spin->si_rem_accents = sal_to_bool(items[2]);
|
|
|
4811 else
|
|
|
4812 /* when "to" is "_" it means empty */
|
|
|
4813 add_fromto(spin, &spin->si_sal, items[1],
|
|
|
4814 STRCMP(items[2], "_") == 0 ? (char_u *)""
|
|
|
4815 : items[2]);
|
|
|
4816 }
|
|
236
|
4817 }
|
|
375
|
4818 else if (STRCMP(items[0], "SOFOFROM") == 0 && itemcnt == 2
|
|
|
4819 && (!do_sofo || spin->si_sofofr == NULL))
|
|
|
4820 {
|
|
|
4821 if (do_sofo)
|
|
|
4822 spin->si_sofofr = vim_strsave(items[1]);
|
|
|
4823 }
|
|
|
4824 else if (STRCMP(items[0], "SOFOTO") == 0 && itemcnt == 2
|
|
|
4825 && (!do_sofo || spin->si_sofoto == NULL))
|
|
|
4826 {
|
|
|
4827 if (do_sofo)
|
|
|
4828 spin->si_sofoto = vim_strsave(items[1]);
|
|
|
4829 }
|
|
300
|
4830 else
|
|
481
|
4831 smsg((char_u *)_("Unrecognized or duplicate item in %s line %d: %s"),
|
|
236
|
4832 fname, lnum, items[0]);
|
|
|
4833 }
|
|
|
4834 }
|
|
|
4835
|
|
375
|
4836 if (do_sofo && (spin->si_sofofr == NULL) != (spin->si_sofoto == NULL))
|
|
|
4837 smsg((char_u *)_("Missing SOFO%s line in %s"),
|
|
|
4838 spin->si_sofofr == NULL ? "FROM" : "TO", fname);
|
|
|
4839 if (spin->si_sofofr != NULL && spin->si_sal.ga_len > 0)
|
|
|
4840 smsg((char_u *)_("Both SAL and SOFO lines in %s"), fname);
|
|
|
4841
|
|
255
|
4842 if (fol != NULL || low != NULL || upp != NULL)
|
|
|
4843 {
|
|
351
|
4844 if (spin->si_clear_chartab)
|
|
|
4845 {
|
|
|
4846 /* Clear the char type tables, don't want to use any of the
|
|
|
4847 * currently used spell properties. */
|
|
|
4848 init_spell_chartab();
|
|
|
4849 spin->si_clear_chartab = FALSE;
|
|
|
4850 }
|
|
|
4851
|
|
316
|
4852 /*
|
|
|
4853 * Don't write a word table for an ASCII file, so that we don't check
|
|
|
4854 * for conflicts with a word table that matches 'encoding'.
|
|
324
|
4855 * Don't write one for utf-8 either, we use utf_*() and
|
|
316
|
4856 * mb_get_class(), the list of chars in the file will be incomplete.
|
|
|
4857 */
|
|
|
4858 if (!spin->si_ascii
|
|
|
4859 #ifdef FEAT_MBYTE
|
|
|
4860 && !enc_utf8
|
|
|
4861 #endif
|
|
|
4862 )
|
|
260
|
4863 {
|
|
|
4864 if (fol == NULL || low == NULL || upp == NULL)
|
|
|
4865 smsg((char_u *)_("Missing FOL/LOW/UPP line in %s"), fname);
|
|
|
4866 else
|
|
316
|
4867 (void)set_spell_chartab(fol, low, upp);
|
|
260
|
4868 }
|
|
255
|
4869
|
|
|
4870 vim_free(fol);
|
|
|
4871 vim_free(low);
|
|
|
4872 vim_free(upp);
|
|
|
4873 }
|
|
|
4874
|
|
481
|
4875 /* Use compound specifications of the .aff file for the spell info. */
|
|
483
|
4876 if (aff->af_compmax != 0)
|
|
|
4877 {
|
|
|
4878 if (spin->si_compmax != 0 && spin->si_compmax != aff->af_compmax)
|
|
|
4879 smsg((char_u *)_("COMPOUNDMAX value differs from what is used in another .aff file"));
|
|
|
4880 else
|
|
|
4881 spin->si_compmax = aff->af_compmax;
|
|
|
4882 }
|
|
|
4883
|
|
481
|
4884 if (aff->af_compminlen != 0)
|
|
|
4885 {
|
|
|
4886 if (spin->si_compminlen != 0
|
|
|
4887 && spin->si_compminlen != aff->af_compminlen)
|
|
|
4888 smsg((char_u *)_("COMPOUNDMIN value differs from what is used in another .aff file"));
|
|
|
4889 else
|
|
|
4890 spin->si_compminlen = aff->af_compminlen;
|
|
|
4891 }
|
|
|
4892
|
|
483
|
4893 if (aff->af_compsylmax != 0)
|
|
|
4894 {
|
|
|
4895 if (aff->af_syllable == NULL)
|
|
|
4896 smsg((char_u *)_("COMPOUNDSYLMAX without SYLLABLE"));
|
|
|
4897
|
|
|
4898 if (spin->si_compsylmax != 0
|
|
|
4899 && spin->si_compsylmax != aff->af_compsylmax)
|
|
|
4900 smsg((char_u *)_("COMPOUNDSYLMAX value differs from what is used in another .aff file"));
|
|
|
4901 else
|
|
|
4902 spin->si_compsylmax = aff->af_compsylmax;
|
|
|
4903 }
|
|
|
4904
|
|
481
|
4905 if (aff->af_compflags != NULL)
|
|
|
4906 {
|
|
|
4907 if (spin->si_compflags != NULL
|
|
|
4908 && STRCMP(spin->si_compflags, aff->af_compflags) != 0)
|
|
|
4909 smsg((char_u *)_("COMPOUNDFLAG(S) value differs from what is used in another .aff file"));
|
|
|
4910 else
|
|
|
4911 spin->si_compflags = aff->af_compflags;
|
|
|
4912 }
|
|
|
4913
|
|
483
|
4914 if (aff->af_syllable != NULL)
|
|
|
4915 {
|
|
|
4916 if (spin->si_syllable != NULL
|
|
|
4917 && STRCMP(spin->si_syllable, aff->af_syllable) != 0)
|
|
|
4918 smsg((char_u *)_("SYLLABLE value differs from what is used in another .aff file"));
|
|
|
4919 else
|
|
|
4920 spin->si_syllable = aff->af_syllable;
|
|
|
4921 }
|
|
|
4922
|
|
236
|
4923 vim_free(pc);
|
|
|
4924 fclose(fd);
|
|
|
4925 return aff;
|
|
|
4926 }
|
|
|
4927
|
|
|
4928 /*
|
|
339
|
4929 * Return TRUE if strings "s1" and "s2" are equal. Also consider both being
|
|
|
4930 * NULL as equal.
|
|
|
4931 */
|
|
|
4932 static int
|
|
|
4933 str_equal(s1, s2)
|
|
|
4934 char_u *s1;
|
|
|
4935 char_u *s2;
|
|
|
4936 {
|
|
|
4937 if (s1 == NULL || s2 == NULL)
|
|
|
4938 return s1 == s2;
|
|
|
4939 return STRCMP(s1, s2) == 0;
|
|
|
4940 }
|
|
|
4941
|
|
|
4942 /*
|
|
323
|
4943 * Add a from-to item to "gap". Used for REP and SAL items.
|
|
|
4944 * They are stored case-folded.
|
|
|
4945 */
|
|
|
4946 static void
|
|
|
4947 add_fromto(spin, gap, from, to)
|
|
|
4948 spellinfo_T *spin;
|
|
|
4949 garray_T *gap;
|
|
|
4950 char_u *from;
|
|
|
4951 char_u *to;
|
|
|
4952 {
|
|
|
4953 fromto_T *ftp;
|
|
|
4954 char_u word[MAXWLEN];
|
|
|
4955
|
|
|
4956 if (ga_grow(gap, 1) == OK)
|
|
|
4957 {
|
|
|
4958 ftp = ((fromto_T *)gap->ga_data) + gap->ga_len;
|
|
|
4959 (void)spell_casefold(from, STRLEN(from), word, MAXWLEN);
|
|
474
|
4960 ftp->ft_from = getroom_save(spin, word);
|
|
323
|
4961 (void)spell_casefold(to, STRLEN(to), word, MAXWLEN);
|
|
474
|
4962 ftp->ft_to = getroom_save(spin, word);
|
|
323
|
4963 ++gap->ga_len;
|
|
|
4964 }
|
|
|
4965 }
|
|
|
4966
|
|
|
4967 /*
|
|
|
4968 * Convert a boolean argument in a SAL line to TRUE or FALSE;
|
|
|
4969 */
|
|
|
4970 static int
|
|
|
4971 sal_to_bool(s)
|
|
|
4972 char_u *s;
|
|
|
4973 {
|
|
|
4974 return STRCMP(s, "1") == 0 || STRCMP(s, "true") == 0;
|
|
|
4975 }
|
|
|
4976
|
|
|
4977 /*
|
|
240
|
4978 * Return TRUE if string "s" contains a non-ASCII character (128 or higher).
|
|
|
4979 * When "s" is NULL FALSE is returned.
|
|
|
4980 */
|
|
|
4981 static int
|
|
|
4982 has_non_ascii(s)
|
|
|
4983 char_u *s;
|
|
|
4984 {
|
|
|
4985 char_u *p;
|
|
|
4986
|
|
|
4987 if (s != NULL)
|
|
|
4988 for (p = s; *p != NUL; ++p)
|
|
|
4989 if (*p >= 128)
|
|
|
4990 return TRUE;
|
|
|
4991 return FALSE;
|
|
|
4992 }
|
|
|
4993
|
|
|
4994 /*
|
|
236
|
4995 * Free the structure filled by spell_read_aff().
|
|
|
4996 */
|
|
|
4997 static void
|
|
|
4998 spell_free_aff(aff)
|
|
|
4999 afffile_T *aff;
|
|
|
5000 {
|
|
|
5001 hashtab_T *ht;
|
|
|
5002 hashitem_T *hi;
|
|
|
5003 int todo;
|
|
|
5004 affheader_T *ah;
|
|
300
|
5005 affentry_T *ae;
|
|
236
|
5006
|
|
|
5007 vim_free(aff->af_enc);
|
|
|
5008
|
|
339
|
5009 /* All this trouble to free the "ae_prog" items... */
|
|
236
|
5010 for (ht = &aff->af_pref; ; ht = &aff->af_suff)
|
|
|
5011 {
|
|
|
5012 todo = ht->ht_used;
|
|
|
5013 for (hi = ht->ht_array; todo > 0; ++hi)
|
|
|
5014 {
|
|
|
5015 if (!HASHITEM_EMPTY(hi))
|
|
|
5016 {
|
|
|
5017 --todo;
|
|
|
5018 ah = HI2AH(hi);
|
|
300
|
5019 for (ae = ah->ah_first; ae != NULL; ae = ae->ae_next)
|
|
|
5020 vim_free(ae->ae_prog);
|
|
236
|
5021 }
|
|
|
5022 }
|
|
|
5023 if (ht == &aff->af_suff)
|
|
|
5024 break;
|
|
|
5025 }
|
|
300
|
5026
|
|
236
|
5027 hash_clear(&aff->af_pref);
|
|
|
5028 hash_clear(&aff->af_suff);
|
|
|
5029 }
|
|
|
5030
|
|
|
5031 /*
|
|
300
|
5032 * Read dictionary file "fname".
|
|
236
|
5033 * Returns OK or FAIL;
|
|
|
5034 */
|
|
|
5035 static int
|
|
474
|
5036 spell_read_dic(spin, fname, affile)
|
|
|
5037 spellinfo_T *spin;
|
|
236
|
5038 char_u *fname;
|
|
300
|
5039 afffile_T *affile;
|
|
236
|
5040 {
|
|
300
|
5041 hashtab_T ht;
|
|
236
|
5042 char_u line[MAXLINELEN];
|
|
481
|
5043 char_u *p;
|
|
300
|
5044 char_u *afflist;
|
|
483
|
5045 char_u store_afflist[MAXWLEN];
|
|
|
5046 int pfxlen;
|
|
|
5047 int need_affix;
|
|
300
|
5048 char_u *dw;
|
|
236
|
5049 char_u *pc;
|
|
|
5050 char_u *w;
|
|
|
5051 int l;
|
|
|
5052 hash_T hash;
|
|
|
5053 hashitem_T *hi;
|
|
|
5054 FILE *fd;
|
|
|
5055 int lnum = 1;
|
|
300
|
5056 int non_ascii = 0;
|
|
|
5057 int retval = OK;
|
|
|
5058 char_u message[MAXLINELEN + MAXWLEN];
|
|
307
|
5059 int flags;
|
|
481
|
5060 int duplicate = 0;
|
|
236
|
5061
|
|
300
|
5062 /*
|
|
|
5063 * Open the file.
|
|
|
5064 */
|
|
310
|
5065 fd = mch_fopen((char *)fname, "r");
|
|
236
|
5066 if (fd == NULL)
|
|
|
5067 {
|
|
|
5068 EMSG2(_(e_notopen), fname);
|
|
|
5069 return FAIL;
|
|
|
5070 }
|
|
|
5071
|
|
300
|
5072 /* The hashtable is only used to detect duplicated words. */
|
|
|
5073 hash_init(&ht);
|
|
|
5074
|
|
310
|
5075 if (spin->si_verbose || p_verbose > 2)
|
|
|
5076 {
|
|
|
5077 if (!spin->si_verbose)
|
|
|
5078 verbose_enter();
|
|
366
|
5079 smsg((char_u *)_("Reading dictionary file %s ..."), fname);
|
|
310
|
5080 out_flush();
|
|
|
5081 if (!spin->si_verbose)
|
|
|
5082 verbose_leave();
|
|
|
5083 }
|
|
236
|
5084
|
|
474
|
5085 /* start with a message for the first line */
|
|
|
5086 spin->si_msg_count = 999999;
|
|
|
5087
|
|
236
|
5088 /* Read and ignore the first line: word count. */
|
|
|
5089 (void)vim_fgets(line, MAXLINELEN, fd);
|
|
324
|
5090 if (!vim_isdigit(*skipwhite(line)))
|
|
236
|
5091 EMSG2(_("E760: No word count in %s"), fname);
|
|
|
5092
|
|
|
5093 /*
|
|
|
5094 * Read all the lines in the file one by one.
|
|
|
5095 * The words are converted to 'encoding' here, before being added to
|
|
|
5096 * the hashtable.
|
|
|
5097 */
|
|
255
|
5098 while (!vim_fgets(line, MAXLINELEN, fd) && !got_int)
|
|
236
|
5099 {
|
|
255
|
5100 line_breakcheck();
|
|
236
|
5101 ++lnum;
|
|
455
|
5102 if (line[0] == '#' || line[0] == '/')
|
|
351
|
5103 continue; /* comment line */
|
|
236
|
5104
|
|
300
|
5105 /* Remove CR, LF and white space from the end. White space halfway
|
|
|
5106 * the word is kept to allow e.g., "et al.". */
|
|
236
|
5107 l = STRLEN(line);
|
|
|
5108 while (l > 0 && line[l - 1] <= ' ')
|
|
|
5109 --l;
|
|
|
5110 if (l == 0)
|
|
|
5111 continue; /* empty line */
|
|
|
5112 line[l] = NUL;
|
|
|
5113
|
|
481
|
5114 /* Find the optional affix names. Replace the SLASH character by a
|
|
|
5115 * slash. */
|
|
|
5116 afflist = NULL;
|
|
|
5117 for (p = line; *p != NUL; mb_ptr_adv(p))
|
|
|
5118 {
|
|
|
5119 if (*p == affile->af_slash)
|
|
|
5120 *p = '/';
|
|
|
5121 else if (*p == '/')
|
|
|
5122 {
|
|
|
5123 *p = NUL;
|
|
|
5124 afflist = p + 1;
|
|
|
5125 break;
|
|
|
5126 }
|
|
|
5127 }
|
|
236
|
5128
|
|
300
|
5129 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */
|
|
|
5130 if (spin->si_ascii && has_non_ascii(line))
|
|
|
5131 {
|
|
|
5132 ++non_ascii;
|
|
240
|
5133 continue;
|
|
300
|
5134 }
|
|
240
|
5135
|
|
310
|
5136 #ifdef FEAT_MBYTE
|
|
236
|
5137 /* Convert from "SET" to 'encoding' when needed. */
|
|
300
|
5138 if (spin->si_conv.vc_type != CONV_NONE)
|
|
236
|
5139 {
|
|
300
|
5140 pc = string_convert(&spin->si_conv, line, NULL);
|
|
255
|
5141 if (pc == NULL)
|
|
|
5142 {
|
|
|
5143 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
5144 fname, lnum, line);
|
|
|
5145 continue;
|
|
|
5146 }
|
|
236
|
5147 w = pc;
|
|
|
5148 }
|
|
|
5149 else
|
|
310
|
5150 #endif
|
|
236
|
5151 {
|
|
|
5152 pc = NULL;
|
|
|
5153 w = line;
|
|
|
5154 }
|
|
|
5155
|
|
474
|
5156 /* This takes time, print a message every 10000 words. */
|
|
|
5157 if (spin->si_verbose && spin->si_msg_count > 10000)
|
|
|
5158 {
|
|
|
5159 spin->si_msg_count = 0;
|
|
339
|
5160 vim_snprintf((char *)message, sizeof(message),
|
|
|
5161 _("line %6d, word %6d - %s"),
|
|
|
5162 lnum, spin->si_foldwcount + spin->si_keepwcount, w);
|
|
|
5163 msg_start();
|
|
|
5164 msg_puts_long_attr(message, 0);
|
|
|
5165 msg_clr_eos();
|
|
|
5166 msg_didout = FALSE;
|
|
|
5167 msg_col = 0;
|
|
|
5168 out_flush();
|
|
|
5169 }
|
|
|
5170
|
|
300
|
5171 /* Store the word in the hashtable to be able to find duplicates. */
|
|
474
|
5172 dw = (char_u *)getroom_save(spin, w);
|
|
236
|
5173 if (dw == NULL)
|
|
300
|
5174 retval = FAIL;
|
|
|
5175 vim_free(pc);
|
|
|
5176 if (retval == FAIL)
|
|
236
|
5177 break;
|
|
|
5178
|
|
300
|
5179 hash = hash_hash(dw);
|
|
|
5180 hi = hash_lookup(&ht, dw, hash);
|
|
236
|
5181 if (!HASHITEM_EMPTY(hi))
|
|
481
|
5182 {
|
|
|
5183 if (p_verbose > 0)
|
|
|
5184 smsg((char_u *)_("Duplicate word in %s line %d: %s"),
|
|
375
|
5185 fname, lnum, dw);
|
|
481
|
5186 else if (duplicate == 0)
|
|
|
5187 smsg((char_u *)_("First duplicate word in %s line %d: %s"),
|
|
|
5188 fname, lnum, dw);
|
|
|
5189 ++duplicate;
|
|
|
5190 }
|
|
236
|
5191 else
|
|
300
|
5192 hash_add_item(&ht, hi, dw, hash);
|
|
|
5193
|
|
307
|
5194 flags = 0;
|
|
483
|
5195 store_afflist[0] = NUL;
|
|
|
5196 pfxlen = 0;
|
|
|
5197 need_affix = FALSE;
|
|
307
|
5198 if (afflist != NULL)
|
|
|
5199 {
|
|
|
5200 /* Check for affix name that stands for keep-case word and stands
|
|
|
5201 * for rare word (if defined). */
|
|
310
|
5202 if (affile->af_kep != NUL
|
|
|
5203 && vim_strchr(afflist, affile->af_kep) != NULL)
|
|
389
|
5204 flags |= WF_KEEPCAP | WF_FIXCAP;
|
|
307
|
5205 if (affile->af_rar != NUL
|
|
|
5206 && vim_strchr(afflist, affile->af_rar) != NULL)
|
|
|
5207 flags |= WF_RARE;
|
|
346
|
5208 if (affile->af_bad != NUL
|
|
|
5209 && vim_strchr(afflist, affile->af_bad) != NULL)
|
|
|
5210 flags |= WF_BANNED;
|
|
483
|
5211 if (affile->af_needaffix != NUL
|
|
|
5212 && vim_strchr(afflist, affile->af_needaffix) != NULL)
|
|
|
5213 need_affix = TRUE;
|
|
339
|
5214
|
|
|
5215 if (affile->af_pfxpostpone)
|
|
|
5216 /* Need to store the list of prefix IDs with the word. */
|
|
483
|
5217 pfxlen = get_pfxlist(affile, afflist, store_afflist);
|
|
|
5218
|
|
|
5219 if (spin->si_compflags != NULL)
|
|
|
5220 /* Need to store the list of compound flags with the word.
|
|
|
5221 * Concatenate them to the list of prefix IDs. */
|
|
|
5222 get_compflags(spin, afflist, store_afflist + pfxlen);
|
|
307
|
5223 }
|
|
|
5224
|
|
300
|
5225 /* Add the word to the word tree(s). */
|
|
483
|
5226 if (store_word(spin, dw, flags, spin->si_region,
|
|
|
5227 store_afflist, need_affix) == FAIL)
|
|
300
|
5228 retval = FAIL;
|
|
236
|
5229
|
|
300
|
5230 if (afflist != NULL)
|
|
|
5231 {
|
|
|
5232 /* Find all matching suffixes and add the resulting words.
|
|
|
5233 * Additionally do matching prefixes that combine. */
|
|
474
|
5234 if (store_aff_word(spin, dw, afflist, affile,
|
|
307
|
5235 &affile->af_suff, &affile->af_pref,
|
|
483
|
5236 FALSE, flags, store_afflist, pfxlen) == FAIL)
|
|
300
|
5237 retval = FAIL;
|
|
|
5238
|
|
|
5239 /* Find all matching prefixes and add the resulting words. */
|
|
474
|
5240 if (store_aff_word(spin, dw, afflist, affile,
|
|
339
|
5241 &affile->af_pref, NULL,
|
|
483
|
5242 FALSE, flags, store_afflist, pfxlen) == FAIL)
|
|
300
|
5243 retval = FAIL;
|
|
|
5244 }
|
|
236
|
5245 }
|
|
|
5246
|
|
481
|
5247 if (duplicate > 0)
|
|
|
5248 smsg((char_u *)_("%d duplicate word(s) in %s"), duplicate, fname);
|
|
300
|
5249 if (spin->si_ascii && non_ascii > 0)
|
|
481
|
5250 smsg((char_u *)_("Ignored %d word(s) with non-ASCII characters in %s"),
|
|
|
5251 non_ascii, fname);
|
|
300
|
5252 hash_clear(&ht);
|
|
|
5253
|
|
236
|
5254 fclose(fd);
|
|
300
|
5255 return retval;
|
|
236
|
5256 }
|
|
|
5257
|
|
|
5258 /*
|
|
339
|
5259 * Get the list of prefix IDs from the affix list "afflist".
|
|
|
5260 * Used for PFXPOSTPONE.
|
|
483
|
5261 * Put the resulting flags in "store_afflist[MAXWLEN]" with a terminating NUL
|
|
|
5262 * and return the number of affixes.
|
|
|
5263 */
|
|
|
5264 static int
|
|
|
5265 get_pfxlist(affile, afflist, store_afflist)
|
|
339
|
5266 afffile_T *affile;
|
|
|
5267 char_u *afflist;
|
|
483
|
5268 char_u *store_afflist;
|
|
339
|
5269 {
|
|
|
5270 char_u *p;
|
|
483
|
5271 int cnt = 0;
|
|
339
|
5272 char_u key[2];
|
|
|
5273 hashitem_T *hi;
|
|
|
5274
|
|
|
5275 key[1] = NUL;
|
|
483
|
5276 for (p = afflist; *p != NUL; ++p)
|
|
|
5277 {
|
|
|
5278 key[0] = *p;
|
|
|
5279 hi = hash_find(&affile->af_pref, key);
|
|
|
5280 if (!HASHITEM_EMPTY(hi))
|
|
|
5281 /* This is a prefix ID, use the new number. */
|
|
|
5282 store_afflist[cnt++] = HI2AH(hi)->ah_newID;
|
|
|
5283 }
|
|
|
5284
|
|
|
5285 store_afflist[cnt] = NUL;
|
|
|
5286 return cnt;
|
|
339
|
5287 }
|
|
|
5288
|
|
|
5289 /*
|
|
481
|
5290 * Get the list of affix IDs from the affix list "afflist" that are used for
|
|
|
5291 * compound words.
|
|
483
|
5292 * Puts the flags in "store_afflist[]".
|
|
|
5293 */
|
|
|
5294 static void
|
|
|
5295 get_compflags(spin, afflist, store_afflist)
|
|
481
|
5296 spellinfo_T *spin;
|
|
|
5297 char_u *afflist;
|
|
483
|
5298 char_u *store_afflist;
|
|
481
|
5299 {
|
|
|
5300 char_u *p;
|
|
483
|
5301 int cnt = 0;
|
|
|
5302
|
|
|
5303 for (p = afflist; *p != NUL; ++p)
|
|
|
5304 /* A flag is a compound flag if it appears in "si_compflags" and
|
|
|
5305 * it's not a special character. */
|
|
|
5306 if (vim_strchr(spin->si_compflags, *p) != NULL
|
|
|
5307 && vim_strchr((char_u *)"+*[]/", *p) == NULL)
|
|
|
5308 store_afflist[cnt++] = *p;
|
|
|
5309
|
|
|
5310 store_afflist[cnt] = NUL;
|
|
481
|
5311 }
|
|
|
5312
|
|
|
5313 /*
|
|
300
|
5314 * Apply affixes to a word and store the resulting words.
|
|
|
5315 * "ht" is the hashtable with affentry_T that need to be applied, either
|
|
|
5316 * prefixes or suffixes.
|
|
|
5317 * "xht", when not NULL, is the prefix hashtable, to be used additionally on
|
|
|
5318 * the resulting words for combining affixes.
|
|
|
5319 *
|
|
|
5320 * Returns FAIL when out of memory.
|
|
236
|
5321 */
|
|
300
|
5322 static int
|
|
483
|
5323 store_aff_word(spin, word, afflist, affile, ht, xht, comb, flags,
|
|
|
5324 pfxlist, pfxlen)
|
|
474
|
5325 spellinfo_T *spin; /* spell info */
|
|
300
|
5326 char_u *word; /* basic word start */
|
|
|
5327 char_u *afflist; /* list of names of supported affixes */
|
|
339
|
5328 afffile_T *affile;
|
|
300
|
5329 hashtab_T *ht;
|
|
|
5330 hashtab_T *xht;
|
|
|
5331 int comb; /* only use affixes that combine */
|
|
307
|
5332 int flags; /* flags for the word */
|
|
339
|
5333 char_u *pfxlist; /* list of prefix IDs */
|
|
483
|
5334 int pfxlen; /* nr of flags in "pfxlist" for prefixes, rest
|
|
|
5335 * is compound flags */
|
|
236
|
5336 {
|
|
|
5337 int todo;
|
|
|
5338 hashitem_T *hi;
|
|
300
|
5339 affheader_T *ah;
|
|
|
5340 affentry_T *ae;
|
|
|
5341 regmatch_T regmatch;
|
|
|
5342 char_u newword[MAXWLEN];
|
|
|
5343 int retval = OK;
|
|
|
5344 int i;
|
|
|
5345 char_u *p;
|
|
366
|
5346 int use_flags;
|
|
390
|
5347 char_u *use_pfxlist;
|
|
483
|
5348 char_u pfx_pfxlist[MAXWLEN];
|
|
455
|
5349 int c;
|
|
483
|
5350 size_t wordlen = STRLEN(word);
|
|
236
|
5351
|
|
300
|
5352 todo = ht->ht_used;
|
|
|
5353 for (hi = ht->ht_array; todo > 0 && retval == OK; ++hi)
|
|
236
|
5354 {
|
|
|
5355 if (!HASHITEM_EMPTY(hi))
|
|
|
5356 {
|
|
|
5357 --todo;
|
|
300
|
5358 ah = HI2AH(hi);
|
|
236
|
5359
|
|
300
|
5360 /* Check that the affix combines, if required, and that the word
|
|
|
5361 * supports this affix. */
|
|
455
|
5362 c = PTR2CHAR(ah->ah_key);
|
|
|
5363 if ((!comb || ah->ah_combine) && vim_strchr(afflist, c) != NULL)
|
|
236
|
5364 {
|
|
300
|
5365 /* Loop over all affix entries with this name. */
|
|
|
5366 for (ae = ah->ah_first; ae != NULL; ae = ae->ae_next)
|
|
236
|
5367 {
|
|
300
|
5368 /* Check the condition. It's not logical to match case
|
|
|
5369 * here, but it is required for compatibility with
|
|
339
|
5370 * Myspell.
|
|
481
|
5371 * Another requirement from Myspell is that the chop
|
|
|
5372 * string is shorter than the word itself.
|
|
339
|
5373 * For prefixes, when "PFXPOSTPONE" was used, only do
|
|
|
5374 * prefixes with a chop string. */
|
|
300
|
5375 regmatch.regprog = ae->ae_prog;
|
|
|
5376 regmatch.rm_ic = FALSE;
|
|
339
|
5377 if ((xht != NULL || !affile->af_pfxpostpone
|
|
|
5378 || ae->ae_chop != NULL)
|
|
481
|
5379 && (ae->ae_chop == NULL
|
|
|
5380 || STRLEN(ae->ae_chop) < wordlen)
|
|
339
|
5381 && (ae->ae_prog == NULL
|
|
|
5382 || vim_regexec(®match, word, (colnr_T)0)))
|
|
300
|
5383 {
|
|
|
5384 /* Match. Remove the chop and add the affix. */
|
|
|
5385 if (xht == NULL)
|
|
240
|
5386 {
|
|
300
|
5387 /* prefix: chop/add at the start of the word */
|
|
|
5388 if (ae->ae_add == NULL)
|
|
|
5389 *newword = NUL;
|
|
|
5390 else
|
|
|
5391 STRCPY(newword, ae->ae_add);
|
|
|
5392 p = word;
|
|
|
5393 if (ae->ae_chop != NULL)
|
|
310
|
5394 {
|
|
300
|
5395 /* Skip chop string. */
|
|
310
|
5396 #ifdef FEAT_MBYTE
|
|
|
5397 if (has_mbyte)
|
|
324
|
5398 {
|
|
310
|
5399 i = mb_charlen(ae->ae_chop);
|
|
324
|
5400 for ( ; i > 0; --i)
|
|
|
5401 mb_ptr_adv(p);
|
|
|
5402 }
|
|
310
|
5403 else
|
|
|
5404 #endif
|
|
324
|
5405 p += STRLEN(ae->ae_chop);
|
|
310
|
5406 }
|
|
300
|
5407 STRCAT(newword, p);
|
|
|
5408 }
|
|
|
5409 else
|
|
|
5410 {
|
|
|
5411 /* suffix: chop/add at the end of the word */
|
|
|
5412 STRCPY(newword, word);
|
|
|
5413 if (ae->ae_chop != NULL)
|
|
|
5414 {
|
|
|
5415 /* Remove chop string. */
|
|
|
5416 p = newword + STRLEN(newword);
|
|
376
|
5417 i = MB_CHARLEN(ae->ae_chop);
|
|
310
|
5418 for ( ; i > 0; --i)
|
|
300
|
5419 mb_ptr_back(newword, p);
|
|
|
5420 *p = NUL;
|
|
|
5421 }
|
|
|
5422 if (ae->ae_add != NULL)
|
|
|
5423 STRCAT(newword, ae->ae_add);
|
|
240
|
5424 }
|
|
|
5425
|
|
366
|
5426 /* Obey the "rare" flag of the affix. */
|
|
|
5427 if (ae->ae_rare)
|
|
|
5428 use_flags = flags | WF_RARE;
|
|
|
5429 else
|
|
|
5430 use_flags = flags;
|
|
483
|
5431
|
|
|
5432 /* Obey the "nocomp" flag of the affix: don't use the
|
|
|
5433 * compound flags. */
|
|
390
|
5434 use_pfxlist = pfxlist;
|
|
483
|
5435 if (ae->ae_nocomp && pfxlist != NULL)
|
|
|
5436 {
|
|
|
5437 vim_strncpy(pfx_pfxlist, pfxlist, pfxlen);
|
|
|
5438 use_pfxlist = pfx_pfxlist;
|
|
|
5439 }
|
|
390
|
5440
|
|
|
5441 /* When there are postponed prefixes... */
|
|
393
|
5442 if (spin->si_prefroot != NULL
|
|
|
5443 && spin->si_prefroot->wn_sibling != NULL)
|
|
390
|
5444 {
|
|
|
5445 /* ... add a flag to indicate an affix was used. */
|
|
|
5446 use_flags |= WF_HAS_AFF;
|
|
|
5447
|
|
|
5448 /* ... don't use a prefix list if combining
|
|
483
|
5449 * affixes is not allowed. But do use the
|
|
|
5450 * compound flags after them. */
|
|
|
5451 if ((!ah->ah_combine || comb) && pfxlist != NULL)
|
|
|
5452 use_pfxlist += pfxlen;
|
|
390
|
5453 }
|
|
366
|
5454
|
|
300
|
5455 /* Store the modified word. */
|
|
474
|
5456 if (store_word(spin, newword, use_flags,
|
|
483
|
5457 spin->si_region, use_pfxlist, FALSE) == FAIL)
|
|
300
|
5458 retval = FAIL;
|
|
236
|
5459
|
|
300
|
5460 /* When added a suffix and combining is allowed also
|
|
|
5461 * try adding prefixes additionally. */
|
|
|
5462 if (xht != NULL && ah->ah_combine)
|
|
474
|
5463 if (store_aff_word(spin, newword, afflist, affile,
|
|
390
|
5464 xht, NULL, TRUE,
|
|
483
|
5465 use_flags, use_pfxlist, pfxlen) == FAIL)
|
|
300
|
5466 retval = FAIL;
|
|
236
|
5467 }
|
|
|
5468 }
|
|
|
5469 }
|
|
|
5470 }
|
|
|
5471 }
|
|
|
5472
|
|
|
5473 return retval;
|
|
|
5474 }
|
|
|
5475
|
|
|
5476 /*
|
|
300
|
5477 * Read a file with a list of words.
|
|
236
|
5478 */
|
|
|
5479 static int
|
|
474
|
5480 spell_read_wordfile(spin, fname)
|
|
|
5481 spellinfo_T *spin;
|
|
300
|
5482 char_u *fname;
|
|
236
|
5483 {
|
|
300
|
5484 FILE *fd;
|
|
|
5485 long lnum = 0;
|
|
|
5486 char_u rline[MAXLINELEN];
|
|
|
5487 char_u *line;
|
|
|
5488 char_u *pc = NULL;
|
|
381
|
5489 char_u *p;
|
|
300
|
5490 int l;
|
|
|
5491 int retval = OK;
|
|
|
5492 int did_word = FALSE;
|
|
|
5493 int non_ascii = 0;
|
|
307
|
5494 int flags;
|
|
316
|
5495 int regionmask;
|
|
236
|
5496
|
|
300
|
5497 /*
|
|
|
5498 * Open the file.
|
|
|
5499 */
|
|
310
|
5500 fd = mch_fopen((char *)fname, "r");
|
|
300
|
5501 if (fd == NULL)
|
|
236
|
5502 {
|
|
300
|
5503 EMSG2(_(e_notopen), fname);
|
|
|
5504 return FAIL;
|
|
236
|
5505 }
|
|
|
5506
|
|
310
|
5507 if (spin->si_verbose || p_verbose > 2)
|
|
|
5508 {
|
|
|
5509 if (!spin->si_verbose)
|
|
|
5510 verbose_enter();
|
|
366
|
5511 smsg((char_u *)_("Reading word file %s ..."), fname);
|
|
310
|
5512 out_flush();
|
|
|
5513 if (!spin->si_verbose)
|
|
|
5514 verbose_leave();
|
|
|
5515 }
|
|
300
|
5516
|
|
|
5517 /*
|
|
|
5518 * Read all the lines in the file one by one.
|
|
|
5519 */
|
|
|
5520 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int)
|
|
|
5521 {
|
|
|
5522 line_breakcheck();
|
|
|
5523 ++lnum;
|
|
|
5524
|
|
|
5525 /* Skip comment lines. */
|
|
|
5526 if (*rline == '#')
|
|
|
5527 continue;
|
|
|
5528
|
|
|
5529 /* Remove CR, LF and white space from the end. */
|
|
|
5530 l = STRLEN(rline);
|
|
|
5531 while (l > 0 && rline[l - 1] <= ' ')
|
|
|
5532 --l;
|
|
|
5533 if (l == 0)
|
|
|
5534 continue; /* empty or blank line */
|
|
|
5535 rline[l] = NUL;
|
|
|
5536
|
|
|
5537 /* Convert from "=encoding={encoding}" to 'encoding' when needed. */
|
|
|
5538 vim_free(pc);
|
|
310
|
5539 #ifdef FEAT_MBYTE
|
|
300
|
5540 if (spin->si_conv.vc_type != CONV_NONE)
|
|
|
5541 {
|
|
|
5542 pc = string_convert(&spin->si_conv, rline, NULL);
|
|
|
5543 if (pc == NULL)
|
|
|
5544 {
|
|
|
5545 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
5546 fname, lnum, rline);
|
|
|
5547 continue;
|
|
|
5548 }
|
|
|
5549 line = pc;
|
|
|
5550 }
|
|
|
5551 else
|
|
310
|
5552 #endif
|
|
300
|
5553 {
|
|
|
5554 pc = NULL;
|
|
|
5555 line = rline;
|
|
|
5556 }
|
|
|
5557
|
|
307
|
5558 if (*line == '/')
|
|
300
|
5559 {
|
|
307
|
5560 ++line;
|
|
|
5561 if (STRNCMP(line, "encoding=", 9) == 0)
|
|
300
|
5562 {
|
|
|
5563 if (spin->si_conv.vc_type != CONV_NONE)
|
|
316
|
5564 smsg((char_u *)_("Duplicate /encoding= line ignored in %s line %d: %s"),
|
|
|
5565 fname, lnum, line - 1);
|
|
300
|
5566 else if (did_word)
|
|
316
|
5567 smsg((char_u *)_("/encoding= line after word ignored in %s line %d: %s"),
|
|
|
5568 fname, lnum, line - 1);
|
|
300
|
5569 else
|
|
|
5570 {
|
|
310
|
5571 #ifdef FEAT_MBYTE
|
|
|
5572 char_u *enc;
|
|
|
5573
|
|
300
|
5574 /* Setup for conversion to 'encoding'. */
|
|
316
|
5575 line += 10;
|
|
|
5576 enc = enc_canonize(line);
|
|
300
|
5577 if (enc != NULL && !spin->si_ascii
|
|
|
5578 && convert_setup(&spin->si_conv, enc,
|
|
|
5579 p_enc) == FAIL)
|
|
|
5580 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"),
|
|
316
|
5581 fname, line, p_enc);
|
|
300
|
5582 vim_free(enc);
|
|
375
|
5583 spin->si_conv.vc_fail = TRUE;
|
|
310
|
5584 #else
|
|
|
5585 smsg((char_u *)_("Conversion in %s not supported"), fname);
|
|
|
5586 #endif
|
|
300
|
5587 }
|
|
307
|
5588 continue;
|
|
300
|
5589 }
|
|
307
|
5590
|
|
316
|
5591 if (STRNCMP(line, "regions=", 8) == 0)
|
|
|
5592 {
|
|
|
5593 if (spin->si_region_count > 1)
|
|
|
5594 smsg((char_u *)_("Duplicate /regions= line ignored in %s line %d: %s"),
|
|
|
5595 fname, lnum, line);
|
|
|
5596 else
|
|
|
5597 {
|
|
|
5598 line += 8;
|
|
|
5599 if (STRLEN(line) > 16)
|
|
|
5600 smsg((char_u *)_("Too many regions in %s line %d: %s"),
|
|
|
5601 fname, lnum, line);
|
|
|
5602 else
|
|
|
5603 {
|
|
|
5604 spin->si_region_count = STRLEN(line) / 2;
|
|
|
5605 STRCPY(spin->si_region_name, line);
|
|
389
|
5606
|
|
|
5607 /* Adjust the mask for a word valid in all regions. */
|
|
|
5608 spin->si_region = (1 << spin->si_region_count) - 1;
|
|
316
|
5609 }
|
|
|
5610 }
|
|
|
5611 continue;
|
|
|
5612 }
|
|
|
5613
|
|
381
|
5614 smsg((char_u *)_("/ line ignored in %s line %d: %s"),
|
|
|
5615 fname, lnum, line - 1);
|
|
|
5616 continue;
|
|
|
5617 }
|
|
|
5618
|
|
|
5619 flags = 0;
|
|
|
5620 regionmask = spin->si_region;
|
|
|
5621
|
|
|
5622 /* Check for flags and region after a slash. */
|
|
|
5623 p = vim_strchr(line, '/');
|
|
|
5624 if (p != NULL)
|
|
|
5625 {
|
|
|
5626 *p++ = NUL;
|
|
|
5627 while (*p != NUL)
|
|
|
5628 {
|
|
|
5629 if (*p == '=') /* keep-case word */
|
|
389
|
5630 flags |= WF_KEEPCAP | WF_FIXCAP;
|
|
381
|
5631 else if (*p == '!') /* Bad, bad, wicked word. */
|
|
|
5632 flags |= WF_BANNED;
|
|
|
5633 else if (*p == '?') /* Rare word. */
|
|
|
5634 flags |= WF_RARE;
|
|
|
5635 else if (VIM_ISDIGIT(*p)) /* region number(s) */
|
|
316
|
5636 {
|
|
381
|
5637 if ((flags & WF_REGION) == 0) /* first one */
|
|
|
5638 regionmask = 0;
|
|
|
5639 flags |= WF_REGION;
|
|
|
5640
|
|
|
5641 l = *p - '0';
|
|
316
|
5642 if (l > spin->si_region_count)
|
|
|
5643 {
|
|
|
5644 smsg((char_u *)_("Invalid region nr in %s line %d: %s"),
|
|
381
|
5645 fname, lnum, p);
|
|
316
|
5646 break;
|
|
|
5647 }
|
|
|
5648 regionmask |= 1 << (l - 1);
|
|
|
5649 }
|
|
381
|
5650 else
|
|
|
5651 {
|
|
|
5652 smsg((char_u *)_("Unrecognized flags in %s line %d: %s"),
|
|
|
5653 fname, lnum, p);
|
|
|
5654 break;
|
|
|
5655 }
|
|
|
5656 ++p;
|
|
307
|
5657 }
|
|
300
|
5658 }
|
|
|
5659
|
|
|
5660 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */
|
|
|
5661 if (spin->si_ascii && has_non_ascii(line))
|
|
|
5662 {
|
|
|
5663 ++non_ascii;
|
|
|
5664 continue;
|
|
|
5665 }
|
|
|
5666
|
|
|
5667 /* Normal word: store it. */
|
|
483
|
5668 if (store_word(spin, line, flags, regionmask, NULL, FALSE) == FAIL)
|
|
300
|
5669 {
|
|
|
5670 retval = FAIL;
|
|
|
5671 break;
|
|
|
5672 }
|
|
|
5673 did_word = TRUE;
|
|
|
5674 }
|
|
|
5675
|
|
|
5676 vim_free(pc);
|
|
|
5677 fclose(fd);
|
|
|
5678
|
|
310
|
5679 if (spin->si_ascii && non_ascii > 0 && (spin->si_verbose || p_verbose > 2))
|
|
|
5680 {
|
|
|
5681 if (p_verbose > 2)
|
|
|
5682 verbose_enter();
|
|
300
|
5683 smsg((char_u *)_("Ignored %d words with non-ASCII characters"),
|
|
|
5684 non_ascii);
|
|
310
|
5685 if (p_verbose > 2)
|
|
|
5686 verbose_leave();
|
|
|
5687 }
|
|
300
|
5688 return retval;
|
|
236
|
5689 }
|
|
|
5690
|
|
|
5691 /*
|
|
300
|
5692 * Get part of an sblock_T, "len" bytes long.
|
|
474
|
5693 * This avoids calling free() for every little struct we use (and keeping
|
|
|
5694 * track of them).
|
|
300
|
5695 * The memory is cleared to all zeros.
|
|
|
5696 * Returns NULL when out of memory.
|
|
|
5697 */
|
|
|
5698 static void *
|
|
474
|
5699 getroom(spin, len, align)
|
|
|
5700 spellinfo_T *spin;
|
|
447
|
5701 size_t len; /* length needed */
|
|
|
5702 int align; /* align for pointer */
|
|
300
|
5703 {
|
|
|
5704 char_u *p;
|
|
474
|
5705 sblock_T *bl = spin->si_blocks;
|
|
300
|
5706
|
|
447
|
5707 if (align && bl != NULL)
|
|
|
5708 /* Round size up for alignment. On some systems structures need to be
|
|
|
5709 * aligned to the size of a pointer (e.g., SPARC). */
|
|
|
5710 bl->sb_used = (bl->sb_used + sizeof(char *) - 1)
|
|
|
5711 & ~(sizeof(char *) - 1);
|
|
|
5712
|
|
300
|
5713 if (bl == NULL || bl->sb_used + len > SBLOCKSIZE)
|
|
|
5714 {
|
|
|
5715 /* Allocate a block of memory. This is not freed until much later. */
|
|
|
5716 bl = (sblock_T *)alloc_clear((unsigned)(sizeof(sblock_T) + SBLOCKSIZE));
|
|
|
5717 if (bl == NULL)
|
|
|
5718 return NULL;
|
|
474
|
5719 bl->sb_next = spin->si_blocks;
|
|
|
5720 spin->si_blocks = bl;
|
|
300
|
5721 bl->sb_used = 0;
|
|
482
|
5722 ++spin->si_blocks_cnt;
|
|
300
|
5723 }
|
|
|
5724
|
|
|
5725 p = bl->sb_data + bl->sb_used;
|
|
|
5726 bl->sb_used += len;
|
|
|
5727
|
|
|
5728 return p;
|
|
|
5729 }
|
|
|
5730
|
|
|
5731 /*
|
|
|
5732 * Make a copy of a string into memory allocated with getroom().
|
|
|
5733 */
|
|
|
5734 static char_u *
|
|
474
|
5735 getroom_save(spin, s)
|
|
|
5736 spellinfo_T *spin;
|
|
300
|
5737 char_u *s;
|
|
|
5738 {
|
|
|
5739 char_u *sc;
|
|
|
5740
|
|
474
|
5741 sc = (char_u *)getroom(spin, STRLEN(s) + 1, FALSE);
|
|
300
|
5742 if (sc != NULL)
|
|
|
5743 STRCPY(sc, s);
|
|
|
5744 return sc;
|
|
|
5745 }
|
|
|
5746
|
|
|
5747
|
|
|
5748 /*
|
|
|
5749 * Free the list of allocated sblock_T.
|
|
236
|
5750 */
|
|
|
5751 static void
|
|
300
|
5752 free_blocks(bl)
|
|
|
5753 sblock_T *bl;
|
|
236
|
5754 {
|
|
300
|
5755 sblock_T *next;
|
|
236
|
5756
|
|
300
|
5757 while (bl != NULL)
|
|
236
|
5758 {
|
|
300
|
5759 next = bl->sb_next;
|
|
|
5760 vim_free(bl);
|
|
|
5761 bl = next;
|
|
236
|
5762 }
|
|
|
5763 }
|
|
|
5764
|
|
|
5765 /*
|
|
300
|
5766 * Allocate the root of a word tree.
|
|
236
|
5767 */
|
|
300
|
5768 static wordnode_T *
|
|
474
|
5769 wordtree_alloc(spin)
|
|
|
5770 spellinfo_T *spin;
|
|
|
5771 {
|
|
|
5772 return (wordnode_T *)getroom(spin, sizeof(wordnode_T), TRUE);
|
|
236
|
5773 }
|
|
|
5774
|
|
|
5775 /*
|
|
300
|
5776 * Store a word in the tree(s).
|
|
390
|
5777 * Always store it in the case-folded tree. For a keep-case word this is
|
|
|
5778 * useful when the word can also be used with all caps (no WF_FIXCAP flag) and
|
|
|
5779 * used to find suggestions.
|
|
300
|
5780 * For a keep-case word also store it in the keep-case tree.
|
|
482
|
5781 * When "pfxlist" is not NULL store the word for each postponed prefix ID and
|
|
|
5782 * compound flag.
|
|
236
|
5783 */
|
|
|
5784 static int
|
|
483
|
5785 store_word(spin, word, flags, region, pfxlist, need_affix)
|
|
474
|
5786 spellinfo_T *spin;
|
|
300
|
5787 char_u *word;
|
|
307
|
5788 int flags; /* extra flags, WF_BANNED */
|
|
316
|
5789 int region; /* supported region(s) */
|
|
339
|
5790 char_u *pfxlist; /* list of prefix IDs or NULL */
|
|
483
|
5791 int need_affix; /* only store word with affix ID */
|
|
236
|
5792 {
|
|
300
|
5793 int len = STRLEN(word);
|
|
|
5794 int ct = captype(word, word + len);
|
|
|
5795 char_u foldword[MAXWLEN];
|
|
339
|
5796 int res = OK;
|
|
|
5797 char_u *p;
|
|
236
|
5798
|
|
323
|
5799 (void)spell_casefold(word, len, foldword, MAXWLEN);
|
|
339
|
5800 for (p = pfxlist; res == OK; ++p)
|
|
|
5801 {
|
|
483
|
5802 if (!need_affix || (p != NULL && *p != NUL))
|
|
|
5803 res = tree_add_word(spin, foldword, spin->si_foldroot, ct | flags,
|
|
474
|
5804 region, p == NULL ? 0 : *p);
|
|
339
|
5805 if (p == NULL || *p == NUL)
|
|
|
5806 break;
|
|
|
5807 }
|
|
334
|
5808 ++spin->si_foldwcount;
|
|
307
|
5809
|
|
474
|
5810 if (res == OK && (ct == WF_KEEPCAP || (flags & WF_KEEPCAP)))
|
|
334
|
5811 {
|
|
339
|
5812 for (p = pfxlist; res == OK; ++p)
|
|
|
5813 {
|
|
483
|
5814 if (!need_affix || (p != NULL && *p != NUL))
|
|
|
5815 res = tree_add_word(spin, word, spin->si_keeproot, flags,
|
|
474
|
5816 region, p == NULL ? 0 : *p);
|
|
339
|
5817 if (p == NULL || *p == NUL)
|
|
|
5818 break;
|
|
|
5819 }
|
|
334
|
5820 ++spin->si_keepwcount;
|
|
|
5821 }
|
|
300
|
5822 return res;
|
|
236
|
5823 }
|
|
|
5824
|
|
|
5825 /*
|
|
300
|
5826 * Add word "word" to a word tree at "root".
|
|
366
|
5827 * When "flags" < 0 we are adding to the prefix tree where flags is used for
|
|
|
5828 * "rare" and "region" is the condition nr.
|
|
255
|
5829 * Returns FAIL when out of memory.
|
|
236
|
5830 */
|
|
255
|
5831 static int
|
|
481
|
5832 tree_add_word(spin, word, root, flags, region, affixID)
|
|
474
|
5833 spellinfo_T *spin;
|
|
300
|
5834 char_u *word;
|
|
|
5835 wordnode_T *root;
|
|
|
5836 int flags;
|
|
|
5837 int region;
|
|
481
|
5838 int affixID;
|
|
236
|
5839 {
|
|
300
|
5840 wordnode_T *node = root;
|
|
|
5841 wordnode_T *np;
|
|
469
|
5842 wordnode_T *copyp, **copyprev;
|
|
300
|
5843 wordnode_T **prev = NULL;
|
|
|
5844 int i;
|
|
255
|
5845
|
|
300
|
5846 /* Add each byte of the word to the tree, including the NUL at the end. */
|
|
|
5847 for (i = 0; ; ++i)
|
|
255
|
5848 {
|
|
469
|
5849 /* When there is more than one reference to this node we need to make
|
|
|
5850 * a copy, so that we can modify it. Copy the whole list of siblings
|
|
|
5851 * (we don't optimize for a partly shared list of siblings). */
|
|
|
5852 if (node != NULL && node->wn_refs > 1)
|
|
|
5853 {
|
|
|
5854 --node->wn_refs;
|
|
|
5855 copyprev = prev;
|
|
|
5856 for (copyp = node; copyp != NULL; copyp = copyp->wn_sibling)
|
|
|
5857 {
|
|
|
5858 /* Allocate a new node and copy the info. */
|
|
474
|
5859 np = get_wordnode(spin);
|
|
469
|
5860 if (np == NULL)
|
|
|
5861 return FAIL;
|
|
|
5862 np->wn_child = copyp->wn_child;
|
|
|
5863 if (np->wn_child != NULL)
|
|
|
5864 ++np->wn_child->wn_refs; /* child gets extra ref */
|
|
|
5865 np->wn_byte = copyp->wn_byte;
|
|
|
5866 if (np->wn_byte == NUL)
|
|
|
5867 {
|
|
|
5868 np->wn_flags = copyp->wn_flags;
|
|
|
5869 np->wn_region = copyp->wn_region;
|
|
481
|
5870 np->wn_affixID = copyp->wn_affixID;
|
|
469
|
5871 }
|
|
|
5872
|
|
|
5873 /* Link the new node in the list, there will be one ref. */
|
|
|
5874 np->wn_refs = 1;
|
|
|
5875 *copyprev = np;
|
|
|
5876 copyprev = &np->wn_sibling;
|
|
|
5877
|
|
|
5878 /* Let "node" point to the head of the copied list. */
|
|
|
5879 if (copyp == node)
|
|
|
5880 node = np;
|
|
|
5881 }
|
|
|
5882 }
|
|
|
5883
|
|
300
|
5884 /* Look for the sibling that has the same character. They are sorted
|
|
|
5885 * on byte value, thus stop searching when a sibling is found with a
|
|
339
|
5886 * higher byte value. For zero bytes (end of word) the sorting is
|
|
481
|
5887 * done on flags and then on affixID. */
|
|
339
|
5888 while (node != NULL
|
|
|
5889 && (node->wn_byte < word[i]
|
|
|
5890 || (node->wn_byte == NUL
|
|
|
5891 && (flags < 0
|
|
481
|
5892 ? node->wn_affixID < affixID
|
|
390
|
5893 : node->wn_flags < (flags & WN_MASK)
|
|
|
5894 || (node->wn_flags == (flags & WN_MASK)
|
|
481
|
5895 && node->wn_affixID < affixID)))))
|
|
236
|
5896 {
|
|
300
|
5897 prev = &node->wn_sibling;
|
|
|
5898 node = *prev;
|
|
236
|
5899 }
|
|
339
|
5900 if (node == NULL
|
|
|
5901 || node->wn_byte != word[i]
|
|
|
5902 || (word[i] == NUL
|
|
|
5903 && (flags < 0
|
|
390
|
5904 || node->wn_flags != (flags & WN_MASK)
|
|
481
|
5905 || node->wn_affixID != affixID)))
|
|
255
|
5906 {
|
|
300
|
5907 /* Allocate a new node. */
|
|
474
|
5908 np = get_wordnode(spin);
|
|
300
|
5909 if (np == NULL)
|
|
|
5910 return FAIL;
|
|
|
5911 np->wn_byte = word[i];
|
|
469
|
5912
|
|
|
5913 /* If "node" is NULL this is a new child or the end of the sibling
|
|
|
5914 * list: ref count is one. Otherwise use ref count of sibling and
|
|
|
5915 * make ref count of sibling one (matters when inserting in front
|
|
|
5916 * of the list of siblings). */
|
|
|
5917 if (node == NULL)
|
|
|
5918 np->wn_refs = 1;
|
|
|
5919 else
|
|
|
5920 {
|
|
|
5921 np->wn_refs = node->wn_refs;
|
|
|
5922 node->wn_refs = 1;
|
|
|
5923 }
|
|
300
|
5924 *prev = np;
|
|
|
5925 np->wn_sibling = node;
|
|
|
5926 node = np;
|
|
255
|
5927 }
|
|
300
|
5928
|
|
|
5929 if (word[i] == NUL)
|
|
|
5930 {
|
|
|
5931 node->wn_flags = flags;
|
|
|
5932 node->wn_region |= region;
|
|
481
|
5933 node->wn_affixID = affixID;
|
|
300
|
5934 break;
|
|
|
5935 }
|
|
|
5936 prev = &node->wn_child;
|
|
|
5937 node = *prev;
|
|
255
|
5938 }
|
|
469
|
5939 #ifdef SPELL_PRINTTREE
|
|
|
5940 smsg("Added \"%s\"", word);
|
|
|
5941 spell_print_tree(root->wn_sibling);
|
|
|
5942 #endif
|
|
|
5943
|
|
474
|
5944 /* count nr of words added since last message */
|
|
|
5945 ++spin->si_msg_count;
|
|
|
5946
|
|
482
|
5947 if (spin->si_compress_cnt > 1)
|
|
|
5948 {
|
|
|
5949 if (--spin->si_compress_cnt == 1)
|
|
|
5950 /* Did enough words to lower the block count limit. */
|
|
483
|
5951 spin->si_blocks_cnt += compress_inc;
|
|
482
|
5952 }
|
|
|
5953
|
|
469
|
5954 /*
|
|
482
|
5955 * When we have allocated lots of memory we need to compress the word tree
|
|
|
5956 * to free up some room. But compression is slow, and we might actually
|
|
|
5957 * need that room, thus only compress in the following situations:
|
|
|
5958 * 1. When not compressed before (si_compress_cnt == 0): when using
|
|
483
|
5959 * "compress_start" blocks.
|
|
|
5960 * 2. When compressed before and used "compress_inc" blocks before
|
|
|
5961 * adding "compress_added" words (si_compress_cnt > 1).
|
|
|
5962 * 3. When compressed before, added "compress_added" words
|
|
482
|
5963 * (si_compress_cnt == 1) and the number of free nodes drops below the
|
|
|
5964 * maximum word length.
|
|
469
|
5965 */
|
|
482
|
5966 #ifndef SPELL_PRINTTREE
|
|
|
5967 if (spin->si_compress_cnt == 1
|
|
|
5968 ? spin->si_free_count < MAXWLEN
|
|
483
|
5969 : spin->si_blocks_cnt >= compress_start)
|
|
482
|
5970 #endif
|
|
|
5971 {
|
|
|
5972 /* Decrement the block counter. The effect is that we compress again
|
|
483
|
5973 * when the freed up room has been used and another "compress_inc"
|
|
|
5974 * blocks have been allocated. Unless "compress_added" words have
|
|
482
|
5975 * been added, then the limit is put back again. */
|
|
483
|
5976 spin->si_blocks_cnt -= compress_inc;
|
|
|
5977 spin->si_compress_cnt = compress_added;
|
|
482
|
5978
|
|
474
|
5979 if (spin->si_verbose)
|
|
|
5980 {
|
|
|
5981 msg_start();
|
|
|
5982 msg_puts((char_u *)_(msg_compressing));
|
|
|
5983 msg_clr_eos();
|
|
|
5984 msg_didout = FALSE;
|
|
|
5985 msg_col = 0;
|
|
|
5986 out_flush();
|
|
|
5987 }
|
|
482
|
5988
|
|
|
5989 /* Compress both trees. Either they both have many nodes, which makes
|
|
|
5990 * compression useful, or one of them is small, which means
|
|
|
5991 * compression goes fast. */
|
|
|
5992 wordtree_compress(spin, spin->si_foldroot);
|
|
|
5993 wordtree_compress(spin, spin->si_keeproot);
|
|
469
|
5994 }
|
|
255
|
5995
|
|
|
5996 return OK;
|
|
236
|
5997 }
|
|
|
5998
|
|
469
|
5999 /*
|
|
483
|
6000 * Check the 'mkspellmem' option. Return FAIL if it's wrong.
|
|
|
6001 * Sets "sps_flags".
|
|
|
6002 */
|
|
|
6003 int
|
|
|
6004 spell_check_msm()
|
|
|
6005 {
|
|
|
6006 char_u *p = p_msm;
|
|
|
6007 long start = 0;
|
|
|
6008 long inc = 0;
|
|
|
6009 long added = 0;
|
|
|
6010
|
|
|
6011 if (!VIM_ISDIGIT(*p))
|
|
|
6012 return FAIL;
|
|
|
6013 /* block count = (value * 1024) / SBLOCKSIZE (but avoid overflow)*/
|
|
|
6014 start = (getdigits(&p) * 10) / (SBLOCKSIZE / 102);
|
|
|
6015 if (*p != ',')
|
|
|
6016 return FAIL;
|
|
|
6017 ++p;
|
|
|
6018 if (!VIM_ISDIGIT(*p))
|
|
|
6019 return FAIL;
|
|
|
6020 inc = (getdigits(&p) * 102) / (SBLOCKSIZE / 10);
|
|
|
6021 if (*p != ',')
|
|
|
6022 return FAIL;
|
|
|
6023 ++p;
|
|
|
6024 if (!VIM_ISDIGIT(*p))
|
|
|
6025 return FAIL;
|
|
|
6026 added = getdigits(&p) * 1024;
|
|
|
6027 if (*p != NUL)
|
|
|
6028 return FAIL;
|
|
|
6029
|
|
|
6030 if (start == 0 || inc == 0 || added == 0 || inc > start)
|
|
|
6031 return FAIL;
|
|
|
6032
|
|
|
6033 compress_start = start;
|
|
|
6034 compress_inc = inc;
|
|
|
6035 compress_added = added;
|
|
|
6036 return OK;
|
|
|
6037 }
|
|
|
6038
|
|
|
6039
|
|
|
6040 /*
|
|
469
|
6041 * Get a wordnode_T, either from the list of previously freed nodes or
|
|
|
6042 * allocate a new one.
|
|
|
6043 */
|
|
|
6044 static wordnode_T *
|
|
474
|
6045 get_wordnode(spin)
|
|
|
6046 spellinfo_T *spin;
|
|
469
|
6047 {
|
|
|
6048 wordnode_T *n;
|
|
|
6049
|
|
474
|
6050 if (spin->si_first_free == NULL)
|
|
|
6051 n = (wordnode_T *)getroom(spin, sizeof(wordnode_T), TRUE);
|
|
469
|
6052 else
|
|
|
6053 {
|
|
474
|
6054 n = spin->si_first_free;
|
|
|
6055 spin->si_first_free = n->wn_child;
|
|
469
|
6056 vim_memset(n, 0, sizeof(wordnode_T));
|
|
482
|
6057 --spin->si_free_count;
|
|
469
|
6058 }
|
|
|
6059 #ifdef SPELL_PRINTTREE
|
|
474
|
6060 n->wn_nr = ++spin->si_wordnode_nr;
|
|
469
|
6061 #endif
|
|
|
6062 return n;
|
|
|
6063 }
|
|
|
6064
|
|
|
6065 /*
|
|
474
|
6066 * Decrement the reference count on a node (which is the head of a list of
|
|
|
6067 * siblings). If the reference count becomes zero free the node and its
|
|
|
6068 * siblings.
|
|
|
6069 */
|
|
|
6070 static void
|
|
|
6071 deref_wordnode(spin, node)
|
|
|
6072 spellinfo_T *spin;
|
|
|
6073 wordnode_T *node;
|
|
|
6074 {
|
|
|
6075 wordnode_T *np;
|
|
|
6076
|
|
|
6077 if (--node->wn_refs == 0)
|
|
|
6078 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
6079 {
|
|
|
6080 if (np->wn_child != NULL)
|
|
|
6081 deref_wordnode(spin, np->wn_child);
|
|
|
6082 free_wordnode(spin, np);
|
|
|
6083 }
|
|
|
6084 }
|
|
|
6085
|
|
|
6086 /*
|
|
469
|
6087 * Free a wordnode_T for re-use later.
|
|
474
|
6088 * Only the "wn_child" field becomes invalid.
|
|
469
|
6089 */
|
|
|
6090 static void
|
|
474
|
6091 free_wordnode(spin, n)
|
|
|
6092 spellinfo_T *spin;
|
|
|
6093 wordnode_T *n;
|
|
|
6094 {
|
|
|
6095 n->wn_child = spin->si_first_free;
|
|
|
6096 spin->si_first_free = n;
|
|
482
|
6097 ++spin->si_free_count;
|
|
469
|
6098 }
|
|
|
6099
|
|
236
|
6100 /*
|
|
300
|
6101 * Compress a tree: find tails that are identical and can be shared.
|
|
|
6102 */
|
|
|
6103 static void
|
|
474
|
6104 wordtree_compress(spin, root)
|
|
|
6105 spellinfo_T *spin;
|
|
300
|
6106 wordnode_T *root;
|
|
|
6107 {
|
|
|
6108 hashtab_T ht;
|
|
|
6109 int n;
|
|
|
6110 int tot = 0;
|
|
469
|
6111 int perc;
|
|
300
|
6112
|
|
474
|
6113 /* Skip the root itself, it's not actually used. The first sibling is the
|
|
|
6114 * start of the tree. */
|
|
|
6115 if (root->wn_sibling != NULL)
|
|
300
|
6116 {
|
|
|
6117 hash_init(&ht);
|
|
474
|
6118 n = node_compress(spin, root->wn_sibling, &ht, &tot);
|
|
469
|
6119
|
|
|
6120 #ifndef SPELL_PRINTTREE
|
|
310
|
6121 if (spin->si_verbose || p_verbose > 2)
|
|
469
|
6122 #endif
|
|
310
|
6123 {
|
|
|
6124 if (!spin->si_verbose)
|
|
|
6125 verbose_enter();
|
|
469
|
6126 if (tot > 1000000)
|
|
|
6127 perc = (tot - n) / (tot / 100);
|
|
482
|
6128 else if (tot == 0)
|
|
|
6129 perc = 0;
|
|
469
|
6130 else
|
|
|
6131 perc = (tot - n) * 100 / tot;
|
|
310
|
6132 smsg((char_u *)_("Compressed %d of %d nodes; %d%% remaining"),
|
|
469
|
6133 n, tot, perc);
|
|
310
|
6134 if (p_verbose > 2)
|
|
|
6135 verbose_leave();
|
|
|
6136 }
|
|
469
|
6137 #ifdef SPELL_PRINTTREE
|
|
474
|
6138 spell_print_tree(root->wn_sibling);
|
|
469
|
6139 #endif
|
|
300
|
6140 hash_clear(&ht);
|
|
|
6141 }
|
|
|
6142 }
|
|
|
6143
|
|
|
6144 /*
|
|
|
6145 * Compress a node, its siblings and its children, depth first.
|
|
|
6146 * Returns the number of compressed nodes.
|
|
236
|
6147 */
|
|
255
|
6148 static int
|
|
474
|
6149 node_compress(spin, node, ht, tot)
|
|
|
6150 spellinfo_T *spin;
|
|
300
|
6151 wordnode_T *node;
|
|
|
6152 hashtab_T *ht;
|
|
|
6153 int *tot; /* total count of nodes before compressing,
|
|
|
6154 incremented while going through the tree */
|
|
236
|
6155 {
|
|
300
|
6156 wordnode_T *np;
|
|
|
6157 wordnode_T *tp;
|
|
|
6158 wordnode_T *child;
|
|
|
6159 hash_T hash;
|
|
236
|
6160 hashitem_T *hi;
|
|
300
|
6161 int len = 0;
|
|
|
6162 unsigned nr, n;
|
|
|
6163 int compressed = 0;
|
|
236
|
6164
|
|
300
|
6165 /*
|
|
|
6166 * Go through the list of siblings. Compress each child and then try
|
|
|
6167 * finding an identical child to replace it.
|
|
|
6168 * Note that with "child" we mean not just the node that is pointed to,
|
|
|
6169 * but the whole list of siblings, of which the node is the first.
|
|
|
6170 */
|
|
482
|
6171 for (np = node; np != NULL && !got_int; np = np->wn_sibling)
|
|
236
|
6172 {
|
|
300
|
6173 ++len;
|
|
|
6174 if ((child = np->wn_child) != NULL)
|
|
|
6175 {
|
|
346
|
6176 /* Compress the child. This fills hashkey. */
|
|
474
|
6177 compressed += node_compress(spin, child, ht, tot);
|
|
300
|
6178
|
|
|
6179 /* Try to find an identical child. */
|
|
346
|
6180 hash = hash_hash(child->wn_u1.hashkey);
|
|
|
6181 hi = hash_lookup(ht, child->wn_u1.hashkey, hash);
|
|
300
|
6182 tp = NULL;
|
|
|
6183 if (!HASHITEM_EMPTY(hi))
|
|
|
6184 {
|
|
|
6185 /* There are children with an identical hash value. Now check
|
|
|
6186 * if there is one that is really identical. */
|
|
346
|
6187 for (tp = HI2WN(hi); tp != NULL; tp = tp->wn_u2.next)
|
|
300
|
6188 if (node_equal(child, tp))
|
|
|
6189 {
|
|
|
6190 /* Found one! Now use that child in place of the
|
|
469
|
6191 * current one. This means the current child and all
|
|
|
6192 * its siblings is unlinked from the tree. */
|
|
|
6193 ++tp->wn_refs;
|
|
474
|
6194 deref_wordnode(spin, child);
|
|
300
|
6195 np->wn_child = tp;
|
|
|
6196 ++compressed;
|
|
|
6197 break;
|
|
|
6198 }
|
|
|
6199 if (tp == NULL)
|
|
|
6200 {
|
|
|
6201 /* No other child with this hash value equals the child of
|
|
|
6202 * the node, add it to the linked list after the first
|
|
|
6203 * item. */
|
|
|
6204 tp = HI2WN(hi);
|
|
346
|
6205 child->wn_u2.next = tp->wn_u2.next;
|
|
|
6206 tp->wn_u2.next = child;
|
|
300
|
6207 }
|
|
|
6208 }
|
|
|
6209 else
|
|
|
6210 /* No other child has this hash value, add it to the
|
|
|
6211 * hashtable. */
|
|
346
|
6212 hash_add_item(ht, hi, child->wn_u1.hashkey, hash);
|
|
300
|
6213 }
|
|
236
|
6214 }
|
|
300
|
6215 *tot += len;
|
|
|
6216
|
|
|
6217 /*
|
|
|
6218 * Make a hash key for the node and its siblings, so that we can quickly
|
|
|
6219 * find a lookalike node. This must be done after compressing the sibling
|
|
|
6220 * list, otherwise the hash key would become invalid by the compression.
|
|
|
6221 */
|
|
346
|
6222 node->wn_u1.hashkey[0] = len;
|
|
300
|
6223 nr = 0;
|
|
|
6224 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
6225 {
|
|
300
|
6226 if (np->wn_byte == NUL)
|
|
481
|
6227 /* end node: use wn_flags, wn_region and wn_affixID */
|
|
|
6228 n = np->wn_flags + (np->wn_region << 8) + (np->wn_affixID << 16);
|
|
300
|
6229 else
|
|
|
6230 /* byte node: use the byte value and the child pointer */
|
|
|
6231 n = np->wn_byte + ((long_u)np->wn_child << 8);
|
|
|
6232 nr = nr * 101 + n;
|
|
236
|
6233 }
|
|
300
|
6234
|
|
|
6235 /* Avoid NUL bytes, it terminates the hash key. */
|
|
|
6236 n = nr & 0xff;
|
|
346
|
6237 node->wn_u1.hashkey[1] = n == 0 ? 1 : n;
|
|
300
|
6238 n = (nr >> 8) & 0xff;
|
|
346
|
6239 node->wn_u1.hashkey[2] = n == 0 ? 1 : n;
|
|
300
|
6240 n = (nr >> 16) & 0xff;
|
|
346
|
6241 node->wn_u1.hashkey[3] = n == 0 ? 1 : n;
|
|
300
|
6242 n = (nr >> 24) & 0xff;
|
|
346
|
6243 node->wn_u1.hashkey[4] = n == 0 ? 1 : n;
|
|
|
6244 node->wn_u1.hashkey[5] = NUL;
|
|
300
|
6245
|
|
482
|
6246 /* Check for CTRL-C pressed now and then. */
|
|
|
6247 fast_breakcheck();
|
|
|
6248
|
|
300
|
6249 return compressed;
|
|
|
6250 }
|
|
|
6251
|
|
|
6252 /*
|
|
|
6253 * Return TRUE when two nodes have identical siblings and children.
|
|
|
6254 */
|
|
|
6255 static int
|
|
|
6256 node_equal(n1, n2)
|
|
|
6257 wordnode_T *n1;
|
|
|
6258 wordnode_T *n2;
|
|
|
6259 {
|
|
|
6260 wordnode_T *p1;
|
|
|
6261 wordnode_T *p2;
|
|
|
6262
|
|
|
6263 for (p1 = n1, p2 = n2; p1 != NULL && p2 != NULL;
|
|
|
6264 p1 = p1->wn_sibling, p2 = p2->wn_sibling)
|
|
|
6265 if (p1->wn_byte != p2->wn_byte
|
|
|
6266 || (p1->wn_byte == NUL
|
|
|
6267 ? (p1->wn_flags != p2->wn_flags
|
|
339
|
6268 || p1->wn_region != p2->wn_region
|
|
481
|
6269 || p1->wn_affixID != p2->wn_affixID)
|
|
300
|
6270 : (p1->wn_child != p2->wn_child)))
|
|
|
6271 break;
|
|
|
6272
|
|
|
6273 return p1 == NULL && p2 == NULL;
|
|
236
|
6274 }
|
|
|
6275
|
|
|
6276 /*
|
|
|
6277 * Write a number to file "fd", MSB first, in "len" bytes.
|
|
|
6278 */
|
|
255
|
6279 void
|
|
236
|
6280 put_bytes(fd, nr, len)
|
|
|
6281 FILE *fd;
|
|
|
6282 long_u nr;
|
|
|
6283 int len;
|
|
|
6284 {
|
|
|
6285 int i;
|
|
|
6286
|
|
|
6287 for (i = len - 1; i >= 0; --i)
|
|
|
6288 putc((int)(nr >> (i * 8)), fd);
|
|
|
6289 }
|
|
|
6290
|
|
323
|
6291 static int
|
|
|
6292 #ifdef __BORLANDC__
|
|
|
6293 _RTLENTRYF
|
|
|
6294 #endif
|
|
|
6295 rep_compare __ARGS((const void *s1, const void *s2));
|
|
|
6296
|
|
|
6297 /*
|
|
|
6298 * Function given to qsort() to sort the REP items on "from" string.
|
|
|
6299 */
|
|
|
6300 static int
|
|
|
6301 #ifdef __BORLANDC__
|
|
|
6302 _RTLENTRYF
|
|
|
6303 #endif
|
|
|
6304 rep_compare(s1, s2)
|
|
|
6305 const void *s1;
|
|
|
6306 const void *s2;
|
|
|
6307 {
|
|
|
6308 fromto_T *p1 = (fromto_T *)s1;
|
|
|
6309 fromto_T *p2 = (fromto_T *)s2;
|
|
|
6310
|
|
|
6311 return STRCMP(p1->ft_from, p2->ft_from);
|
|
|
6312 }
|
|
|
6313
|
|
236
|
6314 /*
|
|
483
|
6315 * Write the Vim .spl file "fname".
|
|
236
|
6316 */
|
|
|
6317 static void
|
|
474
|
6318 write_vim_spell(spin, fname)
|
|
|
6319 spellinfo_T *spin;
|
|
236
|
6320 char_u *fname;
|
|
|
6321 {
|
|
300
|
6322 FILE *fd;
|
|
|
6323 int regionmask;
|
|
236
|
6324 int round;
|
|
300
|
6325 wordnode_T *tree;
|
|
|
6326 int nodecount;
|
|
323
|
6327 int i;
|
|
|
6328 int l;
|
|
|
6329 garray_T *gap;
|
|
|
6330 fromto_T *ftp;
|
|
|
6331 char_u *p;
|
|
|
6332 int rr;
|
|
236
|
6333
|
|
310
|
6334 fd = mch_fopen((char *)fname, "w");
|
|
300
|
6335 if (fd == NULL)
|
|
236
|
6336 {
|
|
|
6337 EMSG2(_(e_notopen), fname);
|
|
|
6338 return;
|
|
|
6339 }
|
|
|
6340
|
|
483
|
6341 /* <HEADER>: <fileID> <versionnr> */
|
|
300
|
6342 /* <fileID> */
|
|
|
6343 if (fwrite(VIMSPELLMAGIC, VIMSPELLMAGICL, (size_t)1, fd) != 1)
|
|
|
6344 EMSG(_(e_write));
|
|
483
|
6345 putc(VIMSPELLVERSION, fd); /* <versionnr> */
|
|
|
6346
|
|
|
6347 /*
|
|
|
6348 * <SECTIONS>: <section> ... <sectionend>
|
|
|
6349 */
|
|
|
6350
|
|
|
6351 /* SN_REGION: <regionname> ...
|
|
|
6352 * Write the region names only if there is more than one. */
|
|
316
|
6353 if (spin->si_region_count > 1)
|
|
236
|
6354 {
|
|
483
|
6355 putc(SN_REGION, fd); /* <sectionID> */
|
|
|
6356 putc(SNF_REQUIRED, fd); /* <sectionflags> */
|
|
|
6357 l = spin->si_region_count * 2;
|
|
|
6358 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
6359 fwrite(spin->si_region_name, (size_t)l, (size_t)1, fd);
|
|
|
6360 /* <regionname> ... */
|
|
316
|
6361 regionmask = (1 << spin->si_region_count) - 1;
|
|
236
|
6362 }
|
|
|
6363 else
|
|
300
|
6364 regionmask = 0;
|
|
483
|
6365
|
|
|
6366 /* SN_CHARFLAGS: <charflagslen> <charflags> <folcharslen> <folchars>
|
|
|
6367 *
|
|
|
6368 * The table with character flags and the table for case folding.
|
|
|
6369 * This makes sure the same characters are recognized as word characters
|
|
|
6370 * when generating an when using a spell file.
|
|
260
|
6371 * Skip this for ASCII, the table may conflict with the one used for
|
|
323
|
6372 * 'encoding'.
|
|
|
6373 * Also skip this for an .add.spl file, the main spell file must contain
|
|
|
6374 * the table (avoids that it conflicts). File is shorter too.
|
|
|
6375 */
|
|
483
|
6376 if (!spin->si_ascii && !spin->si_add)
|
|
|
6377 {
|
|
|
6378 char_u folchars[128 * 8];
|
|
|
6379 int flags;
|
|
|
6380
|
|
485
|
6381 putc(SN_CHARFLAGS, fd); /* <sectionID> */
|
|
483
|
6382 putc(SNF_REQUIRED, fd); /* <sectionflags> */
|
|
|
6383
|
|
|
6384 /* Form the <folchars> string first, we need to know its length. */
|
|
|
6385 l = 0;
|
|
|
6386 for (i = 128; i < 256; ++i)
|
|
|
6387 {
|
|
|
6388 #ifdef FEAT_MBYTE
|
|
|
6389 if (has_mbyte)
|
|
|
6390 l += mb_char2bytes(spelltab.st_fold[i], folchars + l);
|
|
|
6391 else
|
|
|
6392 #endif
|
|
|
6393 folchars[l++] = spelltab.st_fold[i];
|
|
|
6394 }
|
|
|
6395 put_bytes(fd, (long_u)(1 + 128 + 2 + l), 4); /* <sectionlen> */
|
|
|
6396
|
|
|
6397 fputc(128, fd); /* <charflagslen> */
|
|
|
6398 for (i = 128; i < 256; ++i)
|
|
|
6399 {
|
|
|
6400 flags = 0;
|
|
|
6401 if (spelltab.st_isw[i])
|
|
|
6402 flags |= CF_WORD;
|
|
|
6403 if (spelltab.st_isu[i])
|
|
|
6404 flags |= CF_UPPER;
|
|
|
6405 fputc(flags, fd); /* <charflags> */
|
|
|
6406 }
|
|
|
6407
|
|
|
6408 put_bytes(fd, (long_u)l, 2); /* <folcharslen> */
|
|
|
6409 fwrite(folchars, (size_t)l, (size_t)1, fd); /* <folchars> */
|
|
|
6410 }
|
|
|
6411
|
|
|
6412 /* SN_MIDWORD: <midword> */
|
|
|
6413 if (spin->si_midword != NULL)
|
|
|
6414 {
|
|
|
6415 putc(SN_MIDWORD, fd); /* <sectionID> */
|
|
|
6416 putc(SNF_REQUIRED, fd); /* <sectionflags> */
|
|
|
6417
|
|
366
|
6418 i = STRLEN(spin->si_midword);
|
|
483
|
6419 put_bytes(fd, (long_u)i, 4); /* <sectionlen> */
|
|
366
|
6420 fwrite(spin->si_midword, (size_t)i, (size_t)1, fd); /* <midword> */
|
|
|
6421 }
|
|
|
6422
|
|
483
|
6423 /* SN_PREFCOND: <prefcondcnt> <prefcond> ... */
|
|
|
6424 if (spin->si_prefcond.ga_len > 0)
|
|
|
6425 {
|
|
|
6426 putc(SN_PREFCOND, fd); /* <sectionID> */
|
|
|
6427 putc(SNF_REQUIRED, fd); /* <sectionflags> */
|
|
|
6428
|
|
|
6429 l = write_spell_prefcond(NULL, &spin->si_prefcond);
|
|
|
6430 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
6431
|
|
|
6432 write_spell_prefcond(fd, &spin->si_prefcond);
|
|
|
6433 }
|
|
|
6434
|
|
|
6435 /* SN_REP: <repcount> <rep> ...
|
|
|
6436 * SN_SAL: <salflags> <salcount> <sal> ... */
|
|
366
|
6437
|
|
323
|
6438 /* Sort the REP items. */
|
|
|
6439 qsort(spin->si_rep.ga_data, (size_t)spin->si_rep.ga_len,
|
|
|
6440 sizeof(fromto_T), rep_compare);
|
|
|
6441
|
|
483
|
6442 /* round 1: SN_REP section
|
|
|
6443 * round 2: SN_SAL section (unless SN_SOFO is used) */
|
|
323
|
6444 for (round = 1; round <= 2; ++round)
|
|
|
6445 {
|
|
|
6446 if (round == 1)
|
|
483
|
6447 {
|
|
323
|
6448 gap = &spin->si_rep;
|
|
483
|
6449 putc(SN_REP, fd); /* <sectionID> */
|
|
|
6450 }
|
|
323
|
6451 else
|
|
|
6452 {
|
|
483
|
6453 if (spin->si_sofofr != NULL && spin->si_sofoto != NULL)
|
|
|
6454 /* using SN_SOFO section instead of SN_SAL */
|
|
|
6455 break;
|
|
323
|
6456 gap = &spin->si_sal;
|
|
483
|
6457 putc(SN_SAL, fd); /* <sectionID> */
|
|
|
6458 }
|
|
|
6459
|
|
|
6460 /* This is for making suggestions, section is not required. */
|
|
|
6461 putc(0, fd); /* <sectionflags> */
|
|
|
6462
|
|
|
6463 /* Compute the length of what follows. */
|
|
|
6464 l = 2; /* count <repcount> or <salcount> */
|
|
|
6465 for (i = 0; i < gap->ga_len; ++i)
|
|
|
6466 {
|
|
|
6467 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
6468 l += 1 + STRLEN(ftp->ft_from); /* count <*fromlen> and <*from> */
|
|
|
6469 l += 1 + STRLEN(ftp->ft_to); /* count <*tolen> and <*to> */
|
|
|
6470 }
|
|
|
6471 if (round == 2)
|
|
|
6472 ++l; /* count <salflags> */
|
|
|
6473 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
6474
|
|
|
6475 if (round == 2)
|
|
|
6476 {
|
|
323
|
6477 i = 0;
|
|
|
6478 if (spin->si_followup)
|
|
|
6479 i |= SAL_F0LLOWUP;
|
|
|
6480 if (spin->si_collapse)
|
|
|
6481 i |= SAL_COLLAPSE;
|
|
|
6482 if (spin->si_rem_accents)
|
|
|
6483 i |= SAL_REM_ACCENTS;
|
|
|
6484 putc(i, fd); /* <salflags> */
|
|
|
6485 }
|
|
|
6486
|
|
|
6487 put_bytes(fd, (long_u)gap->ga_len, 2); /* <repcount> or <salcount> */
|
|
|
6488 for (i = 0; i < gap->ga_len; ++i)
|
|
|
6489 {
|
|
|
6490 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */
|
|
|
6491 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */
|
|
|
6492 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
6493 for (rr = 1; rr <= 2; ++rr)
|
|
|
6494 {
|
|
|
6495 p = rr == 1 ? ftp->ft_from : ftp->ft_to;
|
|
|
6496 l = STRLEN(p);
|
|
|
6497 putc(l, fd);
|
|
|
6498 fwrite(p, l, (size_t)1, fd);
|
|
|
6499 }
|
|
|
6500 }
|
|
483
|
6501
|
|
|
6502 }
|
|
|
6503
|
|
|
6504 /* SN_SOFO: <sofofromlen> <sofofrom> <sofotolen> <sofoto>
|
|
|
6505 * This is for making suggestions, section is not required. */
|
|
375
|
6506 if (spin->si_sofofr != NULL && spin->si_sofoto != NULL)
|
|
|
6507 {
|
|
483
|
6508 putc(SN_SOFO, fd); /* <sectionID> */
|
|
|
6509 putc(0, fd); /* <sectionflags> */
|
|
375
|
6510
|
|
|
6511 l = STRLEN(spin->si_sofofr);
|
|
483
|
6512 put_bytes(fd, (long_u)(l + STRLEN(spin->si_sofoto) + 4), 4);
|
|
|
6513 /* <sectionlen> */
|
|
|
6514
|
|
|
6515 put_bytes(fd, (long_u)l, 2); /* <sofofromlen> */
|
|
|
6516 fwrite(spin->si_sofofr, l, (size_t)1, fd); /* <sofofrom> */
|
|
375
|
6517
|
|
|
6518 l = STRLEN(spin->si_sofoto);
|
|
483
|
6519 put_bytes(fd, (long_u)l, 2); /* <sofotolen> */
|
|
|
6520 fwrite(spin->si_sofoto, l, (size_t)1, fd); /* <sofoto> */
|
|
|
6521 }
|
|
|
6522
|
|
|
6523 /* SN_MAP: <mapstr>
|
|
|
6524 * This is for making suggestions, section is not required. */
|
|
|
6525 if (spin->si_map.ga_len > 0)
|
|
|
6526 {
|
|
|
6527 putc(SN_MAP, fd); /* <sectionID> */
|
|
|
6528 putc(0, fd); /* <sectionflags> */
|
|
|
6529 l = spin->si_map.ga_len;
|
|
|
6530 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
6531 fwrite(spin->si_map.ga_data, (size_t)l, (size_t)1, fd);
|
|
|
6532 /* <mapstr> */
|
|
|
6533 }
|
|
|
6534
|
|
|
6535 /* SN_COMPOUND: compound info.
|
|
|
6536 * We don't mark it required, when not supported all compound words will
|
|
|
6537 * be bad words. */
|
|
|
6538 if (spin->si_compflags != NULL)
|
|
|
6539 {
|
|
|
6540 putc(SN_COMPOUND, fd); /* <sectionID> */
|
|
|
6541 putc(0, fd); /* <sectionflags> */
|
|
|
6542
|
|
|
6543 l = STRLEN(spin->si_compflags);
|
|
|
6544 put_bytes(fd, (long_u)(l + 3), 4); /* <sectionlen> */
|
|
|
6545 putc(spin->si_compmax, fd); /* <compmax> */
|
|
|
6546 putc(spin->si_compminlen, fd); /* <compminlen> */
|
|
|
6547 putc(spin->si_compsylmax, fd); /* <compsylmax> */
|
|
|
6548 /* <compflags> */
|
|
|
6549 fwrite(spin->si_compflags, (size_t)l, (size_t)1, fd);
|
|
|
6550 }
|
|
|
6551
|
|
492
|
6552 /* SN_NOBREAK: NOBREAK flag */
|
|
|
6553 if (spin->si_nobreak)
|
|
|
6554 {
|
|
|
6555 putc(SN_NOBREAK, fd); /* <sectionID> */
|
|
|
6556 putc(0, fd); /* <sectionflags> */
|
|
|
6557
|
|
|
6558 /* It's empty, the precense of the section flags the feature. */
|
|
|
6559 put_bytes(fd, (long_u)0, 4); /* <sectionlen> */
|
|
|
6560 }
|
|
|
6561
|
|
483
|
6562 /* SN_SYLLABLE: syllable info.
|
|
|
6563 * We don't mark it required, when not supported syllables will not be
|
|
|
6564 * counted. */
|
|
|
6565 if (spin->si_syllable != NULL)
|
|
|
6566 {
|
|
|
6567 putc(SN_SYLLABLE, fd); /* <sectionID> */
|
|
|
6568 putc(0, fd); /* <sectionflags> */
|
|
|
6569
|
|
|
6570 l = STRLEN(spin->si_syllable);
|
|
|
6571 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
6572 fwrite(spin->si_syllable, (size_t)l, (size_t)1, fd); /* <syllable> */
|
|
|
6573 }
|
|
|
6574
|
|
|
6575 /* end of <SECTIONS> */
|
|
|
6576 putc(SN_END, fd); /* <sectionend> */
|
|
|
6577
|
|
302
|
6578
|
|
236
|
6579 /*
|
|
339
|
6580 * <LWORDTREE> <KWORDTREE> <PREFIXTREE>
|
|
236
|
6581 */
|
|
323
|
6582 spin->si_memtot = 0;
|
|
339
|
6583 for (round = 1; round <= 3; ++round)
|
|
236
|
6584 {
|
|
339
|
6585 if (round == 1)
|
|
469
|
6586 tree = spin->si_foldroot->wn_sibling;
|
|
339
|
6587 else if (round == 2)
|
|
469
|
6588 tree = spin->si_keeproot->wn_sibling;
|
|
339
|
6589 else
|
|
469
|
6590 tree = spin->si_prefroot->wn_sibling;
|
|
236
|
6591
|
|
346
|
6592 /* Clear the index and wnode fields in the tree. */
|
|
|
6593 clear_node(tree);
|
|
|
6594
|
|
300
|
6595 /* Count the number of nodes. Needed to be able to allocate the
|
|
346
|
6596 * memory when reading the nodes. Also fills in index for shared
|
|
300
|
6597 * nodes. */
|
|
346
|
6598 nodecount = put_node(NULL, tree, 0, regionmask, round == 3);
|
|
236
|
6599
|
|
300
|
6600 /* number of nodes in 4 bytes */
|
|
|
6601 put_bytes(fd, (long_u)nodecount, 4); /* <nodecount> */
|
|
302
|
6602 spin->si_memtot += nodecount + nodecount * sizeof(int);
|
|
236
|
6603
|
|
300
|
6604 /* Write the nodes. */
|
|
346
|
6605 (void)put_node(fd, tree, 0, regionmask, round == 3);
|
|
236
|
6606 }
|
|
|
6607
|
|
300
|
6608 fclose(fd);
|
|
236
|
6609 }
|
|
|
6610
|
|
|
6611 /*
|
|
346
|
6612 * Clear the index and wnode fields of "node", it siblings and its
|
|
|
6613 * children. This is needed because they are a union with other items to save
|
|
|
6614 * space.
|
|
|
6615 */
|
|
|
6616 static void
|
|
|
6617 clear_node(node)
|
|
|
6618 wordnode_T *node;
|
|
|
6619 {
|
|
|
6620 wordnode_T *np;
|
|
|
6621
|
|
|
6622 if (node != NULL)
|
|
|
6623 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
6624 {
|
|
|
6625 np->wn_u1.index = 0;
|
|
|
6626 np->wn_u2.wnode = NULL;
|
|
|
6627
|
|
|
6628 if (np->wn_byte != NUL)
|
|
|
6629 clear_node(np->wn_child);
|
|
|
6630 }
|
|
|
6631 }
|
|
|
6632
|
|
|
6633
|
|
|
6634 /*
|
|
300
|
6635 * Dump a word tree at node "node".
|
|
|
6636 *
|
|
|
6637 * This first writes the list of possible bytes (siblings). Then for each
|
|
|
6638 * byte recursively write the children.
|
|
|
6639 *
|
|
|
6640 * NOTE: The code here must match the code in read_tree(), since assumptions
|
|
|
6641 * are made about the indexes (so that we don't have to write them in the
|
|
|
6642 * file).
|
|
236
|
6643 *
|
|
300
|
6644 * Returns the number of nodes used.
|
|
236
|
6645 */
|
|
300
|
6646 static int
|
|
346
|
6647 put_node(fd, node, index, regionmask, prefixtree)
|
|
339
|
6648 FILE *fd; /* NULL when only counting */
|
|
300
|
6649 wordnode_T *node;
|
|
|
6650 int index;
|
|
|
6651 int regionmask;
|
|
339
|
6652 int prefixtree; /* TRUE for PREFIXTREE */
|
|
236
|
6653 {
|
|
300
|
6654 int newindex = index;
|
|
|
6655 int siblingcount = 0;
|
|
|
6656 wordnode_T *np;
|
|
236
|
6657 int flags;
|
|
300
|
6658
|
|
|
6659 /* If "node" is zero the tree is empty. */
|
|
|
6660 if (node == NULL)
|
|
|
6661 return 0;
|
|
|
6662
|
|
|
6663 /* Store the index where this node is written. */
|
|
346
|
6664 node->wn_u1.index = index;
|
|
236
|
6665
|
|
300
|
6666 /* Count the number of siblings. */
|
|
|
6667 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
6668 ++siblingcount;
|
|
236
|
6669
|
|
300
|
6670 /* Write the sibling count. */
|
|
|
6671 if (fd != NULL)
|
|
|
6672 putc(siblingcount, fd); /* <siblingcount> */
|
|
236
|
6673
|
|
300
|
6674 /* Write each sibling byte and optionally extra info. */
|
|
|
6675 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
6676 {
|
|
300
|
6677 if (np->wn_byte == 0)
|
|
|
6678 {
|
|
|
6679 if (fd != NULL)
|
|
|
6680 {
|
|
339
|
6681 /* For a NUL byte (end of word) write the flags etc. */
|
|
|
6682 if (prefixtree)
|
|
300
|
6683 {
|
|
481
|
6684 /* In PREFIXTREE write the required affixID and the
|
|
390
|
6685 * associated condition nr (stored in wn_region). The
|
|
|
6686 * byte value is misused to store the "rare" and "not
|
|
|
6687 * combining" flags */
|
|
455
|
6688 if (np->wn_flags == (short_u)PFX_FLAGS)
|
|
|
6689 putc(BY_NOFLAGS, fd); /* <byte> */
|
|
|
6690 else
|
|
|
6691 {
|
|
390
|
6692 putc(BY_FLAGS, fd); /* <byte> */
|
|
455
|
6693 putc(np->wn_flags, fd); /* <pflags> */
|
|
|
6694 }
|
|
481
|
6695 putc(np->wn_affixID, fd); /* <affixID> */
|
|
339
|
6696 put_bytes(fd, (long_u)np->wn_region, 2); /* <prefcondnr> */
|
|
300
|
6697 }
|
|
|
6698 else
|
|
|
6699 {
|
|
339
|
6700 /* For word trees we write the flag/region items. */
|
|
|
6701 flags = np->wn_flags;
|
|
|
6702 if (regionmask != 0 && np->wn_region != regionmask)
|
|
|
6703 flags |= WF_REGION;
|
|
481
|
6704 if (np->wn_affixID != 0)
|
|
|
6705 flags |= WF_AFX;
|
|
339
|
6706 if (flags == 0)
|
|
|
6707 {
|
|
|
6708 /* word without flags or region */
|
|
|
6709 putc(BY_NOFLAGS, fd); /* <byte> */
|
|
|
6710 }
|
|
|
6711 else
|
|
|
6712 {
|
|
390
|
6713 if (np->wn_flags >= 0x100)
|
|
|
6714 {
|
|
|
6715 putc(BY_FLAGS2, fd); /* <byte> */
|
|
|
6716 putc(flags, fd); /* <flags> */
|
|
|
6717 putc((unsigned)flags >> 8, fd); /* <flags2> */
|
|
|
6718 }
|
|
|
6719 else
|
|
|
6720 {
|
|
|
6721 putc(BY_FLAGS, fd); /* <byte> */
|
|
|
6722 putc(flags, fd); /* <flags> */
|
|
|
6723 }
|
|
339
|
6724 if (flags & WF_REGION)
|
|
|
6725 putc(np->wn_region, fd); /* <region> */
|
|
481
|
6726 if (flags & WF_AFX)
|
|
|
6727 putc(np->wn_affixID, fd); /* <affixID> */
|
|
339
|
6728 }
|
|
300
|
6729 }
|
|
|
6730 }
|
|
|
6731 }
|
|
|
6732 else
|
|
|
6733 {
|
|
346
|
6734 if (np->wn_child->wn_u1.index != 0
|
|
|
6735 && np->wn_child->wn_u2.wnode != node)
|
|
300
|
6736 {
|
|
|
6737 /* The child is written elsewhere, write the reference. */
|
|
|
6738 if (fd != NULL)
|
|
|
6739 {
|
|
|
6740 putc(BY_INDEX, fd); /* <byte> */
|
|
|
6741 /* <nodeidx> */
|
|
346
|
6742 put_bytes(fd, (long_u)np->wn_child->wn_u1.index, 3);
|
|
300
|
6743 }
|
|
|
6744 }
|
|
346
|
6745 else if (np->wn_child->wn_u2.wnode == NULL)
|
|
300
|
6746 /* We will write the child below and give it an index. */
|
|
346
|
6747 np->wn_child->wn_u2.wnode = node;
|
|
236
|
6748
|
|
300
|
6749 if (fd != NULL)
|
|
|
6750 if (putc(np->wn_byte, fd) == EOF) /* <byte> or <xbyte> */
|
|
|
6751 {
|
|
|
6752 EMSG(_(e_write));
|
|
|
6753 return 0;
|
|
|
6754 }
|
|
|
6755 }
|
|
236
|
6756 }
|
|
|
6757
|
|
300
|
6758 /* Space used in the array when reading: one for each sibling and one for
|
|
|
6759 * the count. */
|
|
|
6760 newindex += siblingcount + 1;
|
|
249
|
6761
|
|
300
|
6762 /* Recursively dump the children of each sibling. */
|
|
|
6763 for (np = node; np != NULL; np = np->wn_sibling)
|
|
346
|
6764 if (np->wn_byte != 0 && np->wn_child->wn_u2.wnode == node)
|
|
|
6765 newindex = put_node(fd, np->wn_child, newindex, regionmask,
|
|
339
|
6766 prefixtree);
|
|
249
|
6767
|
|
300
|
6768 return newindex;
|
|
236
|
6769 }
|
|
|
6770
|
|
|
6771
|
|
|
6772 /*
|
|
310
|
6773 * ":mkspell [-ascii] outfile infile ..."
|
|
|
6774 * ":mkspell [-ascii] addfile"
|
|
236
|
6775 */
|
|
|
6776 void
|
|
|
6777 ex_mkspell(eap)
|
|
|
6778 exarg_T *eap;
|
|
|
6779 {
|
|
|
6780 int fcount;
|
|
|
6781 char_u **fnames;
|
|
310
|
6782 char_u *arg = eap->arg;
|
|
|
6783 int ascii = FALSE;
|
|
|
6784
|
|
|
6785 if (STRNCMP(arg, "-ascii", 6) == 0)
|
|
|
6786 {
|
|
|
6787 ascii = TRUE;
|
|
|
6788 arg = skipwhite(arg + 6);
|
|
|
6789 }
|
|
|
6790
|
|
|
6791 /* Expand all the remaining arguments (e.g., $VIMRUNTIME). */
|
|
|
6792 if (get_arglist_exp(arg, &fcount, &fnames) == OK)
|
|
|
6793 {
|
|
323
|
6794 mkspell(fcount, fnames, ascii, eap->forceit, FALSE);
|
|
310
|
6795 FreeWild(fcount, fnames);
|
|
|
6796 }
|
|
|
6797 }
|
|
|
6798
|
|
|
6799 /*
|
|
|
6800 * Create a Vim spell file from one or more word lists.
|
|
|
6801 * "fnames[0]" is the output file name.
|
|
|
6802 * "fnames[fcount - 1]" is the last input file name.
|
|
|
6803 * Exception: when "fnames[0]" ends in ".add" it's used as the input file name
|
|
|
6804 * and ".spl" is appended to make the output file name.
|
|
|
6805 */
|
|
|
6806 static void
|
|
323
|
6807 mkspell(fcount, fnames, ascii, overwrite, added_word)
|
|
310
|
6808 int fcount;
|
|
|
6809 char_u **fnames;
|
|
|
6810 int ascii; /* -ascii argument given */
|
|
|
6811 int overwrite; /* overwrite existing output file */
|
|
323
|
6812 int added_word; /* invoked through "zg" */
|
|
310
|
6813 {
|
|
236
|
6814 char_u fname[MAXPATHL];
|
|
|
6815 char_u wfname[MAXPATHL];
|
|
310
|
6816 char_u **innames;
|
|
|
6817 int incount;
|
|
236
|
6818 afffile_T *(afile[8]);
|
|
|
6819 int i;
|
|
|
6820 int len;
|
|
|
6821 struct stat st;
|
|
255
|
6822 int error = FALSE;
|
|
300
|
6823 spellinfo_T spin;
|
|
|
6824
|
|
|
6825 vim_memset(&spin, 0, sizeof(spin));
|
|
323
|
6826 spin.si_verbose = !added_word;
|
|
310
|
6827 spin.si_ascii = ascii;
|
|
323
|
6828 spin.si_followup = TRUE;
|
|
|
6829 spin.si_rem_accents = TRUE;
|
|
|
6830 ga_init2(&spin.si_rep, (int)sizeof(fromto_T), 20);
|
|
|
6831 ga_init2(&spin.si_sal, (int)sizeof(fromto_T), 20);
|
|
|
6832 ga_init2(&spin.si_map, (int)sizeof(char_u), 100);
|
|
339
|
6833 ga_init2(&spin.si_prefcond, (int)sizeof(char_u *), 50);
|
|
310
|
6834
|
|
|
6835 /* default: fnames[0] is output file, following are input files */
|
|
|
6836 innames = &fnames[1];
|
|
|
6837 incount = fcount - 1;
|
|
|
6838
|
|
|
6839 if (fcount >= 1)
|
|
240
|
6840 {
|
|
310
|
6841 len = STRLEN(fnames[0]);
|
|
|
6842 if (fcount == 1 && len > 4 && STRCMP(fnames[0] + len - 4, ".add") == 0)
|
|
|
6843 {
|
|
|
6844 /* For ":mkspell path/en.latin1.add" output file is
|
|
|
6845 * "path/en.latin1.add.spl". */
|
|
|
6846 innames = &fnames[0];
|
|
|
6847 incount = 1;
|
|
|
6848 vim_snprintf((char *)wfname, sizeof(wfname), "%s.spl", fnames[0]);
|
|
|
6849 }
|
|
366
|
6850 else if (fcount == 1)
|
|
|
6851 {
|
|
|
6852 /* For ":mkspell path/vim" output file is "path/vim.latin1.spl". */
|
|
|
6853 innames = &fnames[0];
|
|
|
6854 incount = 1;
|
|
|
6855 vim_snprintf((char *)wfname, sizeof(wfname), "%s.%s.spl", fnames[0],
|
|
|
6856 spin.si_ascii ? (char_u *)"ascii" : spell_enc());
|
|
|
6857 }
|
|
310
|
6858 else if (len > 4 && STRCMP(fnames[0] + len - 4, ".spl") == 0)
|
|
|
6859 {
|
|
|
6860 /* Name ends in ".spl", use as the file name. */
|
|
323
|
6861 vim_strncpy(wfname, fnames[0], sizeof(wfname) - 1);
|
|
310
|
6862 }
|
|
|
6863 else
|
|
|
6864 /* Name should be language, make the file name from it. */
|
|
|
6865 vim_snprintf((char *)wfname, sizeof(wfname), "%s.%s.spl", fnames[0],
|
|
|
6866 spin.si_ascii ? (char_u *)"ascii" : spell_enc());
|
|
|
6867
|
|
|
6868 /* Check for .ascii.spl. */
|
|
|
6869 if (strstr((char *)gettail(wfname), ".ascii.") != NULL)
|
|
|
6870 spin.si_ascii = TRUE;
|
|
|
6871
|
|
|
6872 /* Check for .add.spl. */
|
|
|
6873 if (strstr((char *)gettail(wfname), ".add.") != NULL)
|
|
|
6874 spin.si_add = TRUE;
|
|
240
|
6875 }
|
|
|
6876
|
|
310
|
6877 if (incount <= 0)
|
|
236
|
6878 EMSG(_(e_invarg)); /* need at least output and input names */
|
|
351
|
6879 else if (vim_strchr(gettail(wfname), '_') != NULL)
|
|
|
6880 EMSG(_("E751: Output file name must not have region name"));
|
|
310
|
6881 else if (incount > 8)
|
|
236
|
6882 EMSG(_("E754: Only up to 8 regions supported"));
|
|
|
6883 else
|
|
|
6884 {
|
|
|
6885 /* Check for overwriting before doing things that may take a lot of
|
|
|
6886 * time. */
|
|
310
|
6887 if (!overwrite && mch_stat((char *)wfname, &st) >= 0)
|
|
236
|
6888 {
|
|
|
6889 EMSG(_(e_exists));
|
|
310
|
6890 return;
|
|
236
|
6891 }
|
|
310
|
6892 if (mch_isdir(wfname))
|
|
236
|
6893 {
|
|
310
|
6894 EMSG2(_(e_isadir2), wfname);
|
|
|
6895 return;
|
|
236
|
6896 }
|
|
|
6897
|
|
|
6898 /*
|
|
|
6899 * Init the aff and dic pointers.
|
|
|
6900 * Get the region names if there are more than 2 arguments.
|
|
|
6901 */
|
|
310
|
6902 for (i = 0; i < incount; ++i)
|
|
236
|
6903 {
|
|
310
|
6904 afile[i] = NULL;
|
|
300
|
6905
|
|
316
|
6906 if (incount > 1)
|
|
236
|
6907 {
|
|
310
|
6908 len = STRLEN(innames[i]);
|
|
|
6909 if (STRLEN(gettail(innames[i])) < 5
|
|
|
6910 || innames[i][len - 3] != '_')
|
|
236
|
6911 {
|
|
310
|
6912 EMSG2(_("E755: Invalid region in %s"), innames[i]);
|
|
|
6913 return;
|
|
236
|
6914 }
|
|
316
|
6915 spin.si_region_name[i * 2] = TOLOWER_ASC(innames[i][len - 2]);
|
|
|
6916 spin.si_region_name[i * 2 + 1] =
|
|
|
6917 TOLOWER_ASC(innames[i][len - 1]);
|
|
236
|
6918 }
|
|
|
6919 }
|
|
316
|
6920 spin.si_region_count = incount;
|
|
236
|
6921
|
|
474
|
6922 spin.si_foldroot = wordtree_alloc(&spin);
|
|
|
6923 spin.si_keeproot = wordtree_alloc(&spin);
|
|
|
6924 spin.si_prefroot = wordtree_alloc(&spin);
|
|
339
|
6925 if (spin.si_foldroot == NULL
|
|
|
6926 || spin.si_keeproot == NULL
|
|
|
6927 || spin.si_prefroot == NULL)
|
|
300
|
6928 {
|
|
469
|
6929 free_blocks(spin.si_blocks);
|
|
310
|
6930 return;
|
|
300
|
6931 }
|
|
|
6932
|
|
351
|
6933 /* When not producing a .add.spl file clear the character table when
|
|
|
6934 * we encounter one in the .aff file. This means we dump the current
|
|
|
6935 * one in the .spl file if the .aff file doesn't define one. That's
|
|
|
6936 * better than guessing the contents, the table will match a
|
|
|
6937 * previously loaded spell file. */
|
|
|
6938 if (!spin.si_add)
|
|
|
6939 spin.si_clear_chartab = TRUE;
|
|
|
6940
|
|
236
|
6941 /*
|
|
|
6942 * Read all the .aff and .dic files.
|
|
|
6943 * Text is converted to 'encoding'.
|
|
300
|
6944 * Words are stored in the case-folded and keep-case trees.
|
|
236
|
6945 */
|
|
310
|
6946 for (i = 0; i < incount && !error; ++i)
|
|
236
|
6947 {
|
|
300
|
6948 spin.si_conv.vc_type = CONV_NONE;
|
|
310
|
6949 spin.si_region = 1 << i;
|
|
|
6950
|
|
|
6951 vim_snprintf((char *)fname, sizeof(fname), "%s.aff", innames[i]);
|
|
300
|
6952 if (mch_stat((char *)fname, &st) >= 0)
|
|
|
6953 {
|
|
|
6954 /* Read the .aff file. Will init "spin->si_conv" based on the
|
|
|
6955 * "SET" line. */
|
|
474
|
6956 afile[i] = spell_read_aff(&spin, fname);
|
|
310
|
6957 if (afile[i] == NULL)
|
|
300
|
6958 error = TRUE;
|
|
|
6959 else
|
|
|
6960 {
|
|
|
6961 /* Read the .dic file and store the words in the trees. */
|
|
|
6962 vim_snprintf((char *)fname, sizeof(fname), "%s.dic",
|
|
310
|
6963 innames[i]);
|
|
474
|
6964 if (spell_read_dic(&spin, fname, afile[i]) == FAIL)
|
|
300
|
6965 error = TRUE;
|
|
|
6966 }
|
|
|
6967 }
|
|
|
6968 else
|
|
|
6969 {
|
|
|
6970 /* No .aff file, try reading the file as a word list. Store
|
|
|
6971 * the words in the trees. */
|
|
474
|
6972 if (spell_read_wordfile(&spin, innames[i]) == FAIL)
|
|
300
|
6973 error = TRUE;
|
|
|
6974 }
|
|
236
|
6975
|
|
310
|
6976 #ifdef FEAT_MBYTE
|
|
236
|
6977 /* Free any conversion stuff. */
|
|
300
|
6978 convert_setup(&spin.si_conv, NULL, NULL);
|
|
310
|
6979 #endif
|
|
236
|
6980 }
|
|
|
6981
|
|
492
|
6982 if (spin.si_compflags != NULL && spin.si_nobreak)
|
|
|
6983 MSG(_("Warning: both compounding and NOBREAK specified"));
|
|
|
6984
|
|
300
|
6985 if (!error)
|
|
236
|
6986 {
|
|
|
6987 /*
|
|
300
|
6988 * Combine tails in the tree.
|
|
236
|
6989 */
|
|
474
|
6990 if (spin.si_verbose || p_verbose > 2)
|
|
|
6991 {
|
|
|
6992 if (!spin.si_verbose)
|
|
310
|
6993 verbose_enter();
|
|
469
|
6994 MSG(_(msg_compressing));
|
|
310
|
6995 out_flush();
|
|
474
|
6996 if (!spin.si_verbose)
|
|
310
|
6997 verbose_leave();
|
|
|
6998 }
|
|
474
|
6999 wordtree_compress(&spin, spin.si_foldroot);
|
|
|
7000 wordtree_compress(&spin, spin.si_keeproot);
|
|
|
7001 wordtree_compress(&spin, spin.si_prefroot);
|
|
236
|
7002 }
|
|
|
7003
|
|
300
|
7004 if (!error)
|
|
|
7005 {
|
|
|
7006 /*
|
|
|
7007 * Write the info in the spell file.
|
|
|
7008 */
|
|
474
|
7009 if (spin.si_verbose || p_verbose > 2)
|
|
|
7010 {
|
|
|
7011 if (!spin.si_verbose)
|
|
310
|
7012 verbose_enter();
|
|
366
|
7013 smsg((char_u *)_("Writing spell file %s ..."), wfname);
|
|
310
|
7014 out_flush();
|
|
474
|
7015 if (!spin.si_verbose)
|
|
310
|
7016 verbose_leave();
|
|
|
7017 }
|
|
|
7018
|
|
474
|
7019 write_vim_spell(&spin, wfname);
|
|
|
7020
|
|
|
7021 if (spin.si_verbose || p_verbose > 2)
|
|
|
7022 {
|
|
|
7023 if (!spin.si_verbose)
|
|
310
|
7024 verbose_enter();
|
|
|
7025 MSG(_("Done!"));
|
|
|
7026 smsg((char_u *)_("Estimated runtime memory use: %d bytes"),
|
|
302
|
7027 spin.si_memtot);
|
|
310
|
7028 out_flush();
|
|
474
|
7029 if (!spin.si_verbose)
|
|
310
|
7030 verbose_leave();
|
|
|
7031 }
|
|
|
7032
|
|
|
7033 /* If the file is loaded need to reload it. */
|
|
323
|
7034 spell_reload_one(wfname, added_word);
|
|
300
|
7035 }
|
|
|
7036
|
|
|
7037 /* Free the allocated memory. */
|
|
323
|
7038 ga_clear(&spin.si_rep);
|
|
|
7039 ga_clear(&spin.si_sal);
|
|
|
7040 ga_clear(&spin.si_map);
|
|
339
|
7041 ga_clear(&spin.si_prefcond);
|
|
366
|
7042 vim_free(spin.si_midword);
|
|
375
|
7043 vim_free(spin.si_sofofr);
|
|
|
7044 vim_free(spin.si_sofoto);
|
|
300
|
7045
|
|
|
7046 /* Free the .aff file structures. */
|
|
310
|
7047 for (i = 0; i < incount; ++i)
|
|
|
7048 if (afile[i] != NULL)
|
|
|
7049 spell_free_aff(afile[i]);
|
|
339
|
7050
|
|
|
7051 /* Free all the bits and pieces at once. */
|
|
|
7052 free_blocks(spin.si_blocks);
|
|
236
|
7053 }
|
|
310
|
7054 }
|
|
|
7055
|
|
|
7056
|
|
|
7057 /*
|
|
385
|
7058 * ":[count]spellgood {word}"
|
|
|
7059 * ":[count]spellwrong {word}"
|
|
310
|
7060 */
|
|
|
7061 void
|
|
|
7062 ex_spell(eap)
|
|
|
7063 exarg_T *eap;
|
|
|
7064 {
|
|
381
|
7065 spell_add_word(eap->arg, STRLEN(eap->arg), eap->cmdidx == CMD_spellwrong,
|
|
385
|
7066 eap->forceit ? 0 : (int)eap->line2);
|
|
236
|
7067 }
|
|
|
7068
|
|
310
|
7069 /*
|
|
|
7070 * Add "word[len]" to 'spellfile' as a good or bad word.
|
|
|
7071 */
|
|
|
7072 void
|
|
385
|
7073 spell_add_word(word, len, bad, index)
|
|
310
|
7074 char_u *word;
|
|
|
7075 int len;
|
|
|
7076 int bad;
|
|
385
|
7077 int index; /* "zG" and "zW": zero, otherwise index in
|
|
|
7078 'spellfile' */
|
|
310
|
7079 {
|
|
|
7080 FILE *fd;
|
|
385
|
7081 buf_T *buf = NULL;
|
|
351
|
7082 int new_spf = FALSE;
|
|
|
7083 struct stat st;
|
|
381
|
7084 char_u *fname;
|
|
385
|
7085 char_u fnamebuf[MAXPATHL];
|
|
|
7086 char_u line[MAXWLEN * 2];
|
|
|
7087 long fpos, fpos_next = 0;
|
|
|
7088 int i;
|
|
|
7089 char_u *spf;
|
|
|
7090
|
|
|
7091 if (index == 0) /* use internal wordlist */
|
|
|
7092 {
|
|
|
7093 if (int_wordlist == NULL)
|
|
|
7094 {
|
|
|
7095 int_wordlist = vim_tempname('s');
|
|
|
7096 if (int_wordlist == NULL)
|
|
381
|
7097 return;
|
|
|
7098 }
|
|
385
|
7099 fname = int_wordlist;
|
|
381
|
7100 }
|
|
310
|
7101 else
|
|
|
7102 {
|
|
381
|
7103 /* If 'spellfile' isn't set figure out a good default value. */
|
|
|
7104 if (*curbuf->b_p_spf == NUL)
|
|
|
7105 {
|
|
|
7106 init_spellfile();
|
|
|
7107 new_spf = TRUE;
|
|
|
7108 }
|
|
|
7109
|
|
|
7110 if (*curbuf->b_p_spf == NUL)
|
|
|
7111 {
|
|
|
7112 EMSG(_("E764: 'spellfile' is not set"));
|
|
|
7113 return;
|
|
|
7114 }
|
|
|
7115
|
|
385
|
7116 for (spf = curbuf->b_p_spf, i = 1; *spf != NUL; ++i)
|
|
|
7117 {
|
|
|
7118 copy_option_part(&spf, fnamebuf, MAXPATHL, ",");
|
|
|
7119 if (i == index)
|
|
|
7120 break;
|
|
|
7121 if (*spf == NUL)
|
|
|
7122 {
|
|
|
7123 EMSGN(_("E765: 'spellfile' does not have %ld enties"), index);
|
|
|
7124 return;
|
|
|
7125 }
|
|
|
7126 }
|
|
|
7127
|
|
310
|
7128 /* Check that the user isn't editing the .add file somewhere. */
|
|
385
|
7129 buf = buflist_findname_exp(fnamebuf);
|
|
310
|
7130 if (buf != NULL && buf->b_ml.ml_mfp == NULL)
|
|
|
7131 buf = NULL;
|
|
|
7132 if (buf != NULL && bufIsChanged(buf))
|
|
381
|
7133 {
|
|
310
|
7134 EMSG(_(e_bufloaded));
|
|
381
|
7135 return;
|
|
|
7136 }
|
|
|
7137
|
|
385
|
7138 fname = fnamebuf;
|
|
|
7139 }
|
|
|
7140
|
|
|
7141 if (bad)
|
|
|
7142 {
|
|
|
7143 /* When the word also appears as good word we need to remove that one,
|
|
|
7144 * since its flags sort before the one with WF_BANNED. */
|
|
|
7145 fd = mch_fopen((char *)fname, "r");
|
|
|
7146 if (fd != NULL)
|
|
|
7147 {
|
|
|
7148 while (!vim_fgets(line, MAXWLEN * 2, fd))
|
|
|
7149 {
|
|
|
7150 fpos = fpos_next;
|
|
|
7151 fpos_next = ftell(fd);
|
|
|
7152 if (STRNCMP(word, line, len) == 0
|
|
|
7153 && (line[len] == '/' || line[len] < ' '))
|
|
|
7154 {
|
|
|
7155 /* Found duplicate word. Remove it by writing a '#' at
|
|
|
7156 * the start of the line. Mixing reading and writing
|
|
|
7157 * doesn't work for all systems, close the file first. */
|
|
|
7158 fclose(fd);
|
|
|
7159 fd = mch_fopen((char *)fname, "r+");
|
|
|
7160 if (fd == NULL)
|
|
|
7161 break;
|
|
|
7162 if (fseek(fd, fpos, SEEK_SET) == 0)
|
|
|
7163 fputc('#', fd);
|
|
|
7164 fseek(fd, fpos_next, SEEK_SET);
|
|
|
7165 }
|
|
|
7166 }
|
|
|
7167 fclose(fd);
|
|
|
7168 }
|
|
381
|
7169 }
|
|
|
7170
|
|
|
7171 fd = mch_fopen((char *)fname, "a");
|
|
|
7172 if (fd == NULL && new_spf)
|
|
|
7173 {
|
|
|
7174 /* We just initialized the 'spellfile' option and can't open the file.
|
|
|
7175 * We may need to create the "spell" directory first. We already
|
|
|
7176 * checked the runtime directory is writable in init_spellfile(). */
|
|
|
7177 STRCPY(NameBuff, fname);
|
|
|
7178 *gettail_sep(NameBuff) = NUL;
|
|
|
7179 if (mch_stat((char *)NameBuff, &st) < 0)
|
|
|
7180 {
|
|
|
7181 /* The directory doesn't exist. Try creating it and opening the
|
|
|
7182 * file again. */
|
|
|
7183 vim_mkdir(NameBuff, 0755);
|
|
|
7184 fd = mch_fopen((char *)fname, "a");
|
|
|
7185 }
|
|
|
7186 }
|
|
|
7187
|
|
|
7188 if (fd == NULL)
|
|
|
7189 EMSG2(_(e_notopen), fname);
|
|
|
7190 else
|
|
|
7191 {
|
|
|
7192 if (bad)
|
|
|
7193 fprintf(fd, "%.*s/!\n", len, word);
|
|
310
|
7194 else
|
|
381
|
7195 fprintf(fd, "%.*s\n", len, word);
|
|
|
7196 fclose(fd);
|
|
|
7197
|
|
|
7198 /* Update the .add.spl file. */
|
|
|
7199 mkspell(1, &fname, FALSE, TRUE, TRUE);
|
|
|
7200
|
|
|
7201 /* If the .add file is edited somewhere, reload it. */
|
|
|
7202 if (buf != NULL)
|
|
|
7203 buf_reload(buf);
|
|
|
7204
|
|
|
7205 redraw_all_later(NOT_VALID);
|
|
310
|
7206 }
|
|
|
7207 }
|
|
|
7208
|
|
|
7209 /*
|
|
|
7210 * Initialize 'spellfile' for the current buffer.
|
|
|
7211 */
|
|
|
7212 static void
|
|
|
7213 init_spellfile()
|
|
|
7214 {
|
|
|
7215 char_u buf[MAXPATHL];
|
|
|
7216 int l;
|
|
|
7217 slang_T *sl;
|
|
|
7218 char_u *rtp;
|
|
351
|
7219 char_u *lend;
|
|
310
|
7220
|
|
|
7221 if (*curbuf->b_p_spl != NUL && curbuf->b_langp.ga_len > 0)
|
|
|
7222 {
|
|
351
|
7223 /* Find the end of the language name. Exclude the region. */
|
|
|
7224 for (lend = curbuf->b_p_spl; *lend != NUL
|
|
|
7225 && vim_strchr((char_u *)",._", *lend) == NULL; ++lend)
|
|
|
7226 ;
|
|
|
7227
|
|
|
7228 /* Loop over all entries in 'runtimepath'. Use the first one where we
|
|
|
7229 * are allowed to write. */
|
|
310
|
7230 rtp = p_rtp;
|
|
|
7231 while (*rtp != NUL)
|
|
|
7232 {
|
|
|
7233 /* Copy the path from 'runtimepath' to buf[]. */
|
|
|
7234 copy_option_part(&rtp, buf, MAXPATHL, ",");
|
|
|
7235 if (filewritable(buf) == 2)
|
|
|
7236 {
|
|
316
|
7237 /* Use the first language name from 'spelllang' and the
|
|
|
7238 * encoding used in the first loaded .spl file. */
|
|
310
|
7239 sl = LANGP_ENTRY(curbuf->b_langp, 0)->lp_slang;
|
|
|
7240 l = STRLEN(buf);
|
|
|
7241 vim_snprintf((char *)buf + l, MAXPATHL - l,
|
|
316
|
7242 "/spell/%.*s.%s.add",
|
|
351
|
7243 (int)(lend - curbuf->b_p_spl), curbuf->b_p_spl,
|
|
310
|
7244 strstr((char *)gettail(sl->sl_fname), ".ascii.") != NULL
|
|
|
7245 ? (char_u *)"ascii" : spell_enc());
|
|
|
7246 set_option_value((char_u *)"spellfile", 0L, buf, OPT_LOCAL);
|
|
|
7247 break;
|
|
|
7248 }
|
|
|
7249 }
|
|
|
7250 }
|
|
|
7251 }
|
|
236
|
7252
|
|
300
|
7253
|
|
307
|
7254 /*
|
|
|
7255 * Init the chartab used for spelling for ASCII.
|
|
|
7256 * EBCDIC is not supported!
|
|
|
7257 */
|
|
|
7258 static void
|
|
|
7259 clear_spell_chartab(sp)
|
|
|
7260 spelltab_T *sp;
|
|
|
7261 {
|
|
324
|
7262 int i;
|
|
307
|
7263
|
|
|
7264 /* Init everything to FALSE. */
|
|
|
7265 vim_memset(sp->st_isw, FALSE, sizeof(sp->st_isw));
|
|
|
7266 vim_memset(sp->st_isu, FALSE, sizeof(sp->st_isu));
|
|
|
7267 for (i = 0; i < 256; ++i)
|
|
324
|
7268 {
|
|
307
|
7269 sp->st_fold[i] = i;
|
|
324
|
7270 sp->st_upper[i] = i;
|
|
|
7271 }
|
|
307
|
7272
|
|
|
7273 /* We include digits. A word shouldn't start with a digit, but handling
|
|
|
7274 * that is done separately. */
|
|
|
7275 for (i = '0'; i <= '9'; ++i)
|
|
|
7276 sp->st_isw[i] = TRUE;
|
|
|
7277 for (i = 'A'; i <= 'Z'; ++i)
|
|
|
7278 {
|
|
|
7279 sp->st_isw[i] = TRUE;
|
|
|
7280 sp->st_isu[i] = TRUE;
|
|
|
7281 sp->st_fold[i] = i + 0x20;
|
|
|
7282 }
|
|
|
7283 for (i = 'a'; i <= 'z'; ++i)
|
|
324
|
7284 {
|
|
307
|
7285 sp->st_isw[i] = TRUE;
|
|
324
|
7286 sp->st_upper[i] = i - 0x20;
|
|
|
7287 }
|
|
307
|
7288 }
|
|
|
7289
|
|
|
7290 /*
|
|
|
7291 * Init the chartab used for spelling. Only depends on 'encoding'.
|
|
|
7292 * Called once while starting up and when 'encoding' changes.
|
|
|
7293 * The default is to use isalpha(), but the spell file should define the word
|
|
|
7294 * characters to make it possible that 'encoding' differs from the current
|
|
390
|
7295 * locale. For utf-8 we don't use isalpha() but our own functions.
|
|
307
|
7296 */
|
|
|
7297 void
|
|
|
7298 init_spell_chartab()
|
|
|
7299 {
|
|
|
7300 int i;
|
|
|
7301
|
|
|
7302 did_set_spelltab = FALSE;
|
|
|
7303 clear_spell_chartab(&spelltab);
|
|
|
7304 #ifdef FEAT_MBYTE
|
|
|
7305 if (enc_dbcs)
|
|
|
7306 {
|
|
|
7307 /* DBCS: assume double-wide characters are word characters. */
|
|
|
7308 for (i = 128; i <= 255; ++i)
|
|
|
7309 if (MB_BYTE2LEN(i) == 2)
|
|
|
7310 spelltab.st_isw[i] = TRUE;
|
|
|
7311 }
|
|
324
|
7312 else if (enc_utf8)
|
|
|
7313 {
|
|
|
7314 for (i = 128; i < 256; ++i)
|
|
|
7315 {
|
|
|
7316 spelltab.st_isu[i] = utf_isupper(i);
|
|
|
7317 spelltab.st_isw[i] = spelltab.st_isu[i] || utf_islower(i);
|
|
|
7318 spelltab.st_fold[i] = utf_fold(i);
|
|
|
7319 spelltab.st_upper[i] = utf_toupper(i);
|
|
|
7320 }
|
|
|
7321 }
|
|
307
|
7322 else
|
|
|
7323 #endif
|
|
|
7324 {
|
|
324
|
7325 /* Rough guess: use locale-dependent library functions. */
|
|
307
|
7326 for (i = 128; i < 256; ++i)
|
|
|
7327 {
|
|
|
7328 if (MB_ISUPPER(i))
|
|
|
7329 {
|
|
324
|
7330 spelltab.st_isw[i] = TRUE;
|
|
307
|
7331 spelltab.st_isu[i] = TRUE;
|
|
|
7332 spelltab.st_fold[i] = MB_TOLOWER(i);
|
|
|
7333 }
|
|
324
|
7334 else if (MB_ISLOWER(i))
|
|
|
7335 {
|
|
|
7336 spelltab.st_isw[i] = TRUE;
|
|
|
7337 spelltab.st_upper[i] = MB_TOUPPER(i);
|
|
|
7338 }
|
|
307
|
7339 }
|
|
|
7340 }
|
|
|
7341 }
|
|
|
7342
|
|
|
7343 static char *e_affform = N_("E761: Format error in affix file FOL, LOW or UPP");
|
|
|
7344 static char *e_affrange = N_("E762: Character in FOL, LOW or UPP is out of range");
|
|
|
7345
|
|
|
7346 /*
|
|
|
7347 * Set the spell character tables from strings in the affix file.
|
|
|
7348 */
|
|
|
7349 static int
|
|
|
7350 set_spell_chartab(fol, low, upp)
|
|
|
7351 char_u *fol;
|
|
|
7352 char_u *low;
|
|
|
7353 char_u *upp;
|
|
|
7354 {
|
|
|
7355 /* We build the new tables here first, so that we can compare with the
|
|
|
7356 * previous one. */
|
|
|
7357 spelltab_T new_st;
|
|
|
7358 char_u *pf = fol, *pl = low, *pu = upp;
|
|
|
7359 int f, l, u;
|
|
|
7360
|
|
|
7361 clear_spell_chartab(&new_st);
|
|
|
7362
|
|
|
7363 while (*pf != NUL)
|
|
|
7364 {
|
|
|
7365 if (*pl == NUL || *pu == NUL)
|
|
|
7366 {
|
|
|
7367 EMSG(_(e_affform));
|
|
|
7368 return FAIL;
|
|
|
7369 }
|
|
|
7370 #ifdef FEAT_MBYTE
|
|
|
7371 f = mb_ptr2char_adv(&pf);
|
|
|
7372 l = mb_ptr2char_adv(&pl);
|
|
|
7373 u = mb_ptr2char_adv(&pu);
|
|
|
7374 #else
|
|
|
7375 f = *pf++;
|
|
|
7376 l = *pl++;
|
|
|
7377 u = *pu++;
|
|
|
7378 #endif
|
|
|
7379 /* Every character that appears is a word character. */
|
|
|
7380 if (f < 256)
|
|
|
7381 new_st.st_isw[f] = TRUE;
|
|
|
7382 if (l < 256)
|
|
|
7383 new_st.st_isw[l] = TRUE;
|
|
|
7384 if (u < 256)
|
|
|
7385 new_st.st_isw[u] = TRUE;
|
|
|
7386
|
|
|
7387 /* if "LOW" and "FOL" are not the same the "LOW" char needs
|
|
|
7388 * case-folding */
|
|
|
7389 if (l < 256 && l != f)
|
|
|
7390 {
|
|
|
7391 if (f >= 256)
|
|
|
7392 {
|
|
|
7393 EMSG(_(e_affrange));
|
|
|
7394 return FAIL;
|
|
|
7395 }
|
|
|
7396 new_st.st_fold[l] = f;
|
|
|
7397 }
|
|
|
7398
|
|
|
7399 /* if "UPP" and "FOL" are not the same the "UPP" char needs
|
|
324
|
7400 * case-folding, it's upper case and the "UPP" is the upper case of
|
|
|
7401 * "FOL" . */
|
|
307
|
7402 if (u < 256 && u != f)
|
|
|
7403 {
|
|
|
7404 if (f >= 256)
|
|
|
7405 {
|
|
|
7406 EMSG(_(e_affrange));
|
|
|
7407 return FAIL;
|
|
|
7408 }
|
|
|
7409 new_st.st_fold[u] = f;
|
|
|
7410 new_st.st_isu[u] = TRUE;
|
|
324
|
7411 new_st.st_upper[f] = u;
|
|
307
|
7412 }
|
|
|
7413 }
|
|
|
7414
|
|
|
7415 if (*pl != NUL || *pu != NUL)
|
|
|
7416 {
|
|
|
7417 EMSG(_(e_affform));
|
|
|
7418 return FAIL;
|
|
|
7419 }
|
|
|
7420
|
|
|
7421 return set_spell_finish(&new_st);
|
|
|
7422 }
|
|
|
7423
|
|
|
7424 /*
|
|
|
7425 * Set the spell character tables from strings in the .spl file.
|
|
|
7426 */
|
|
483
|
7427 static void
|
|
389
|
7428 set_spell_charflags(flags, cnt, fol)
|
|
307
|
7429 char_u *flags;
|
|
389
|
7430 int cnt; /* length of "flags" */
|
|
|
7431 char_u *fol;
|
|
307
|
7432 {
|
|
|
7433 /* We build the new tables here first, so that we can compare with the
|
|
|
7434 * previous one. */
|
|
|
7435 spelltab_T new_st;
|
|
|
7436 int i;
|
|
389
|
7437 char_u *p = fol;
|
|
324
|
7438 int c;
|
|
307
|
7439
|
|
|
7440 clear_spell_chartab(&new_st);
|
|
|
7441
|
|
389
|
7442 for (i = 0; i < 128; ++i)
|
|
|
7443 {
|
|
|
7444 if (i < cnt)
|
|
|
7445 {
|
|
|
7446 new_st.st_isw[i + 128] = (flags[i] & CF_WORD) != 0;
|
|
|
7447 new_st.st_isu[i + 128] = (flags[i] & CF_UPPER) != 0;
|
|
|
7448 }
|
|
|
7449
|
|
|
7450 if (*p != NUL)
|
|
|
7451 {
|
|
307
|
7452 #ifdef FEAT_MBYTE
|
|
389
|
7453 c = mb_ptr2char_adv(&p);
|
|
307
|
7454 #else
|
|
389
|
7455 c = *p++;
|
|
307
|
7456 #endif
|
|
389
|
7457 new_st.st_fold[i + 128] = c;
|
|
|
7458 if (i + 128 != c && new_st.st_isu[i + 128] && c < 256)
|
|
|
7459 new_st.st_upper[c] = i + 128;
|
|
|
7460 }
|
|
307
|
7461 }
|
|
|
7462
|
|
483
|
7463 (void)set_spell_finish(&new_st);
|
|
307
|
7464 }
|
|
|
7465
|
|
|
7466 static int
|
|
|
7467 set_spell_finish(new_st)
|
|
|
7468 spelltab_T *new_st;
|
|
|
7469 {
|
|
|
7470 int i;
|
|
|
7471
|
|
|
7472 if (did_set_spelltab)
|
|
|
7473 {
|
|
|
7474 /* check that it's the same table */
|
|
|
7475 for (i = 0; i < 256; ++i)
|
|
|
7476 {
|
|
|
7477 if (spelltab.st_isw[i] != new_st->st_isw[i]
|
|
|
7478 || spelltab.st_isu[i] != new_st->st_isu[i]
|
|
324
|
7479 || spelltab.st_fold[i] != new_st->st_fold[i]
|
|
|
7480 || spelltab.st_upper[i] != new_st->st_upper[i])
|
|
307
|
7481 {
|
|
|
7482 EMSG(_("E763: Word characters differ between spell files"));
|
|
|
7483 return FAIL;
|
|
|
7484 }
|
|
|
7485 }
|
|
|
7486 }
|
|
|
7487 else
|
|
|
7488 {
|
|
|
7489 /* copy the new spelltab into the one being used */
|
|
|
7490 spelltab = *new_st;
|
|
|
7491 did_set_spelltab = TRUE;
|
|
|
7492 }
|
|
|
7493
|
|
|
7494 return OK;
|
|
|
7495 }
|
|
|
7496
|
|
|
7497 /*
|
|
358
|
7498 * Return TRUE if "p" points to a word character.
|
|
366
|
7499 * As a special case we see "midword" characters as word character when it is
|
|
358
|
7500 * followed by a word character. This finds they'there but not 'they there'.
|
|
366
|
7501 * Thus this only works properly when past the first character of the word.
|
|
358
|
7502 */
|
|
|
7503 static int
|
|
376
|
7504 spell_iswordp(p, buf)
|
|
358
|
7505 char_u *p;
|
|
376
|
7506 buf_T *buf; /* buffer used */
|
|
358
|
7507 {
|
|
366
|
7508 #ifdef FEAT_MBYTE
|
|
358
|
7509 char_u *s;
|
|
366
|
7510 int l;
|
|
|
7511 int c;
|
|
|
7512
|
|
|
7513 if (has_mbyte)
|
|
|
7514 {
|
|
|
7515 l = MB_BYTE2LEN(*p);
|
|
358
|
7516 s = p;
|
|
366
|
7517 if (l == 1)
|
|
|
7518 {
|
|
|
7519 /* be quick for ASCII */
|
|
376
|
7520 if (buf->b_spell_ismw[*p])
|
|
366
|
7521 {
|
|
|
7522 s = p + 1; /* skip a mid-word character */
|
|
|
7523 l = MB_BYTE2LEN(*s);
|
|
|
7524 }
|
|
|
7525 }
|
|
|
7526 else
|
|
|
7527 {
|
|
|
7528 c = mb_ptr2char(p);
|
|
376
|
7529 if (c < 256 ? buf->b_spell_ismw[c]
|
|
|
7530 : (buf->b_spell_ismw_mb != NULL
|
|
|
7531 && vim_strchr(buf->b_spell_ismw_mb, c) != NULL))
|
|
366
|
7532 {
|
|
|
7533 s = p + l;
|
|
|
7534 l = MB_BYTE2LEN(*s);
|
|
|
7535 }
|
|
|
7536 }
|
|
|
7537
|
|
390
|
7538 c = mb_ptr2char(s);
|
|
|
7539 if (c > 255)
|
|
366
|
7540 return mb_get_class(s) >= 2;
|
|
390
|
7541 return spelltab.st_isw[c];
|
|
366
|
7542 }
|
|
358
|
7543 #endif
|
|
366
|
7544
|
|
376
|
7545 return spelltab.st_isw[buf->b_spell_ismw[*p] ? p[1] : p[0]];
|
|
|
7546 }
|
|
|
7547
|
|
|
7548 /*
|
|
|
7549 * Return TRUE if "p" points to a word character.
|
|
|
7550 * Unlike spell_iswordp() this doesn't check for "midword" characters.
|
|
|
7551 */
|
|
|
7552 static int
|
|
|
7553 spell_iswordp_nmw(p)
|
|
|
7554 char_u *p;
|
|
|
7555 {
|
|
|
7556 #ifdef FEAT_MBYTE
|
|
390
|
7557 int c;
|
|
|
7558
|
|
|
7559 if (has_mbyte)
|
|
|
7560 {
|
|
|
7561 c = mb_ptr2char(p);
|
|
|
7562 if (c > 255)
|
|
|
7563 return mb_get_class(p) >= 2;
|
|
|
7564 return spelltab.st_isw[c];
|
|
|
7565 }
|
|
376
|
7566 #endif
|
|
|
7567 return spelltab.st_isw[*p];
|
|
358
|
7568 }
|
|
|
7569
|
|
372
|
7570 #ifdef FEAT_MBYTE
|
|
|
7571 /*
|
|
|
7572 * Return TRUE if "p" points to a word character.
|
|
|
7573 * Wide version of spell_iswordp().
|
|
|
7574 */
|
|
|
7575 static int
|
|
376
|
7576 spell_iswordp_w(p, buf)
|
|
372
|
7577 int *p;
|
|
376
|
7578 buf_T *buf;
|
|
372
|
7579 {
|
|
|
7580 int *s;
|
|
|
7581
|
|
376
|
7582 if (*p < 256 ? buf->b_spell_ismw[*p]
|
|
|
7583 : (buf->b_spell_ismw_mb != NULL
|
|
|
7584 && vim_strchr(buf->b_spell_ismw_mb, *p) != NULL))
|
|
372
|
7585 s = p + 1;
|
|
|
7586 else
|
|
|
7587 s = p;
|
|
|
7588
|
|
390
|
7589 if (*s > 255)
|
|
372
|
7590 {
|
|
|
7591 if (enc_utf8)
|
|
|
7592 return utf_class(*s) >= 2;
|
|
|
7593 if (enc_dbcs)
|
|
|
7594 return dbcs_class((unsigned)*s >> 8, *s & 0xff) >= 2;
|
|
|
7595 return 0;
|
|
|
7596 }
|
|
|
7597 return spelltab.st_isw[*s];
|
|
|
7598 }
|
|
|
7599 #endif
|
|
|
7600
|
|
358
|
7601 /*
|
|
339
|
7602 * Write the table with prefix conditions to the .spl file.
|
|
483
|
7603 * When "fd" is NULL only count the length of what is written.
|
|
|
7604 */
|
|
|
7605 static int
|
|
339
|
7606 write_spell_prefcond(fd, gap)
|
|
|
7607 FILE *fd;
|
|
|
7608 garray_T *gap;
|
|
|
7609 {
|
|
|
7610 int i;
|
|
|
7611 char_u *p;
|
|
|
7612 int len;
|
|
483
|
7613 int totlen;
|
|
|
7614
|
|
|
7615 if (fd != NULL)
|
|
|
7616 put_bytes(fd, (long_u)gap->ga_len, 2); /* <prefcondcnt> */
|
|
|
7617
|
|
|
7618 totlen = 2 + gap->ga_len; /* length of <prefcondcnt> and <condlen> bytes */
|
|
339
|
7619
|
|
|
7620 for (i = 0; i < gap->ga_len; ++i)
|
|
|
7621 {
|
|
|
7622 /* <prefcond> : <condlen> <condstr> */
|
|
|
7623 p = ((char_u **)gap->ga_data)[i];
|
|
483
|
7624 if (p != NULL)
|
|
339
|
7625 {
|
|
|
7626 len = STRLEN(p);
|
|
483
|
7627 if (fd != NULL)
|
|
|
7628 {
|
|
|
7629 fputc(len, fd);
|
|
|
7630 fwrite(p, (size_t)len, (size_t)1, fd);
|
|
|
7631 }
|
|
|
7632 totlen += len;
|
|
|
7633 }
|
|
|
7634 else if (fd != NULL)
|
|
|
7635 fputc(0, fd);
|
|
|
7636 }
|
|
|
7637
|
|
|
7638 return totlen;
|
|
307
|
7639 }
|
|
|
7640
|
|
|
7641 /*
|
|
324
|
7642 * Case-fold "str[len]" into "buf[buflen]". The result is NUL terminated.
|
|
|
7643 * Uses the character definitions from the .spl file.
|
|
307
|
7644 * When using a multi-byte 'encoding' the length may change!
|
|
|
7645 * Returns FAIL when something wrong.
|
|
|
7646 */
|
|
|
7647 static int
|
|
324
|
7648 spell_casefold(str, len, buf, buflen)
|
|
|
7649 char_u *str;
|
|
307
|
7650 int len;
|
|
|
7651 char_u *buf;
|
|
|
7652 int buflen;
|
|
|
7653 {
|
|
|
7654 int i;
|
|
|
7655
|
|
|
7656 if (len >= buflen)
|
|
|
7657 {
|
|
|
7658 buf[0] = NUL;
|
|
|
7659 return FAIL; /* result will not fit */
|
|
|
7660 }
|
|
|
7661
|
|
|
7662 #ifdef FEAT_MBYTE
|
|
|
7663 if (has_mbyte)
|
|
|
7664 {
|
|
324
|
7665 int outi = 0;
|
|
|
7666 char_u *p;
|
|
307
|
7667 int c;
|
|
|
7668
|
|
|
7669 /* Fold one character at a time. */
|
|
324
|
7670 for (p = str; p < str + len; )
|
|
307
|
7671 {
|
|
|
7672 if (outi + MB_MAXBYTES > buflen)
|
|
|
7673 {
|
|
|
7674 buf[outi] = NUL;
|
|
|
7675 return FAIL;
|
|
|
7676 }
|
|
474
|
7677 c = mb_cptr2char_adv(&p);
|
|
324
|
7678 outi += mb_char2bytes(SPELL_TOFOLD(c), buf + outi);
|
|
307
|
7679 }
|
|
|
7680 buf[outi] = NUL;
|
|
|
7681 }
|
|
|
7682 else
|
|
|
7683 #endif
|
|
|
7684 {
|
|
|
7685 /* Be quick for non-multibyte encodings. */
|
|
|
7686 for (i = 0; i < len; ++i)
|
|
324
|
7687 buf[i] = spelltab.st_fold[str[i]];
|
|
307
|
7688 buf[i] = NUL;
|
|
|
7689 }
|
|
|
7690
|
|
|
7691 return OK;
|
|
|
7692 }
|
|
|
7693
|
|
372
|
7694 #define SPS_BEST 1
|
|
|
7695 #define SPS_FAST 2
|
|
|
7696 #define SPS_DOUBLE 4
|
|
|
7697
|
|
|
7698 static int sps_flags = SPS_BEST;
|
|
483
|
7699 static int sps_limit = 9999;
|
|
372
|
7700
|
|
|
7701 /*
|
|
|
7702 * Check the 'spellsuggest' option. Return FAIL if it's wrong.
|
|
483
|
7703 * Sets "sps_flags" and "sps_limit".
|
|
372
|
7704 */
|
|
|
7705 int
|
|
|
7706 spell_check_sps()
|
|
|
7707 {
|
|
|
7708 char_u *p;
|
|
483
|
7709 char_u *s;
|
|
372
|
7710 char_u buf[MAXPATHL];
|
|
|
7711 int f;
|
|
|
7712
|
|
|
7713 sps_flags = 0;
|
|
483
|
7714 sps_limit = 9999;
|
|
372
|
7715
|
|
|
7716 for (p = p_sps; *p != NUL; )
|
|
|
7717 {
|
|
|
7718 copy_option_part(&p, buf, MAXPATHL, ",");
|
|
|
7719
|
|
|
7720 f = 0;
|
|
483
|
7721 if (VIM_ISDIGIT(*buf))
|
|
|
7722 {
|
|
|
7723 s = buf;
|
|
|
7724 sps_limit = getdigits(&s);
|
|
|
7725 if (*s != NUL && !VIM_ISDIGIT(*s))
|
|
|
7726 f = -1;
|
|
|
7727 }
|
|
|
7728 else if (STRCMP(buf, "best") == 0)
|
|
372
|
7729 f = SPS_BEST;
|
|
|
7730 else if (STRCMP(buf, "fast") == 0)
|
|
|
7731 f = SPS_FAST;
|
|
|
7732 else if (STRCMP(buf, "double") == 0)
|
|
|
7733 f = SPS_DOUBLE;
|
|
|
7734 else if (STRNCMP(buf, "expr:", 5) != 0
|
|
|
7735 && STRNCMP(buf, "file:", 5) != 0)
|
|
|
7736 f = -1;
|
|
|
7737
|
|
|
7738 if (f == -1 || (sps_flags != 0 && f != 0))
|
|
|
7739 {
|
|
|
7740 sps_flags = SPS_BEST;
|
|
483
|
7741 sps_limit = 9999;
|
|
372
|
7742 return FAIL;
|
|
|
7743 }
|
|
|
7744 if (f != 0)
|
|
|
7745 sps_flags = f;
|
|
|
7746 }
|
|
|
7747
|
|
|
7748 if (sps_flags == 0)
|
|
|
7749 sps_flags = SPS_BEST;
|
|
|
7750
|
|
|
7751 return OK;
|
|
|
7752 }
|
|
|
7753
|
|
|
7754 /* Remember what "z?" replaced. */
|
|
|
7755 static char_u *repl_from = NULL;
|
|
|
7756 static char_u *repl_to = NULL;
|
|
|
7757
|
|
323
|
7758 /*
|
|
|
7759 * "z?": Find badly spelled word under or after the cursor.
|
|
|
7760 * Give suggestions for the properly spelled word.
|
|
485
|
7761 * When "count" is non-zero use that suggestion.
|
|
323
|
7762 */
|
|
|
7763 void
|
|
485
|
7764 spell_suggest(count)
|
|
|
7765 int count;
|
|
323
|
7766 {
|
|
|
7767 char_u *line;
|
|
|
7768 pos_T prev_cursor = curwin->w_cursor;
|
|
|
7769 char_u wcopy[MAXWLEN + 2];
|
|
|
7770 char_u *p;
|
|
|
7771 int i;
|
|
|
7772 int c;
|
|
|
7773 suginfo_T sug;
|
|
|
7774 suggest_T *stp;
|
|
372
|
7775 int mouse_used;
|
|
388
|
7776 int need_cap;
|
|
483
|
7777 int limit;
|
|
485
|
7778 int selected = count;
|
|
323
|
7779
|
|
344
|
7780 /* Find the start of the badly spelled word. */
|
|
346
|
7781 if (spell_move_to(FORWARD, TRUE, TRUE) == FAIL
|
|
|
7782 || curwin->w_cursor.col > prev_cursor.col)
|
|
|
7783 {
|
|
|
7784 if (!curwin->w_p_spell || *curbuf->b_p_spl == NUL)
|
|
|
7785 return;
|
|
|
7786
|
|
|
7787 /* No bad word or it starts after the cursor: use the word under the
|
|
|
7788 * cursor. */
|
|
|
7789 curwin->w_cursor = prev_cursor;
|
|
|
7790 line = ml_get_curline();
|
|
|
7791 p = line + curwin->w_cursor.col;
|
|
|
7792 /* Backup to before start of word. */
|
|
390
|
7793 while (p > line && spell_iswordp_nmw(p))
|
|
346
|
7794 mb_ptr_back(line, p);
|
|
|
7795 /* Forward to start of word. */
|
|
390
|
7796 while (*p != NUL && !spell_iswordp_nmw(p))
|
|
346
|
7797 mb_ptr_adv(p);
|
|
|
7798
|
|
390
|
7799 if (!spell_iswordp_nmw(p)) /* No word found. */
|
|
346
|
7800 {
|
|
|
7801 beep_flush();
|
|
|
7802 return;
|
|
|
7803 }
|
|
|
7804 curwin->w_cursor.col = p - line;
|
|
323
|
7805 }
|
|
|
7806
|
|
344
|
7807 /* Get the word and its length. */
|
|
|
7808
|
|
388
|
7809 /* Figure out if the word should be capitalised. */
|
|
475
|
7810 need_cap = check_need_cap(curwin->w_cursor.lnum, curwin->w_cursor.col);
|
|
|
7811
|
|
|
7812 line = ml_get_curline();
|
|
388
|
7813
|
|
483
|
7814 /* Get the list of suggestions. Limit to 'lines' - 2 or the number in
|
|
|
7815 * 'spellsuggest', whatever is smaller. */
|
|
|
7816 if (sps_limit > (int)Rows - 2)
|
|
|
7817 limit = (int)Rows - 2;
|
|
|
7818 else
|
|
|
7819 limit = sps_limit;
|
|
|
7820 spell_find_suggest(line + curwin->w_cursor.col, &sug, limit,
|
|
388
|
7821 TRUE, need_cap);
|
|
323
|
7822
|
|
|
7823 if (sug.su_ga.ga_len == 0)
|
|
|
7824 MSG(_("Sorry, no suggestions"));
|
|
485
|
7825 else if (count > 0)
|
|
|
7826 {
|
|
|
7827 if (count > sug.su_ga.ga_len)
|
|
|
7828 smsg((char_u *)_("Sorry, only %ld suggestions"),
|
|
|
7829 (long)sug.su_ga.ga_len);
|
|
|
7830 }
|
|
323
|
7831 else
|
|
|
7832 {
|
|
372
|
7833 vim_free(repl_from);
|
|
|
7834 repl_from = NULL;
|
|
|
7835 vim_free(repl_to);
|
|
|
7836 repl_to = NULL;
|
|
|
7837
|
|
474
|
7838 #ifdef FEAT_RIGHTLEFT
|
|
|
7839 /* When 'rightleft' is set the list is drawn right-left. */
|
|
|
7840 cmdmsg_rl = curwin->w_p_rl;
|
|
|
7841 if (cmdmsg_rl)
|
|
|
7842 msg_col = Columns - 1;
|
|
|
7843 #endif
|
|
|
7844
|
|
323
|
7845 /* List the suggestions. */
|
|
|
7846 msg_start();
|
|
372
|
7847 lines_left = Rows; /* avoid more prompt */
|
|
323
|
7848 vim_snprintf((char *)IObuff, IOSIZE, _("Change \"%.*s\" to:"),
|
|
|
7849 sug.su_badlen, sug.su_badptr);
|
|
474
|
7850 #ifdef FEAT_RIGHTLEFT
|
|
|
7851 if (cmdmsg_rl && STRNCMP(IObuff, "Change", 6) == 0)
|
|
|
7852 {
|
|
|
7853 /* And now the rabbit from the high hat: Avoid showing the
|
|
|
7854 * untranslated message rightleft. */
|
|
|
7855 vim_snprintf((char *)IObuff, IOSIZE, ":ot \"%.*s\" egnahC",
|
|
|
7856 sug.su_badlen, sug.su_badptr);
|
|
|
7857 }
|
|
|
7858 #endif
|
|
323
|
7859 msg_puts(IObuff);
|
|
|
7860 msg_clr_eos();
|
|
|
7861 msg_putchar('\n');
|
|
346
|
7862
|
|
323
|
7863 msg_scroll = TRUE;
|
|
|
7864 for (i = 0; i < sug.su_ga.ga_len; ++i)
|
|
|
7865 {
|
|
344
|
7866 stp = &SUG(sug.su_ga, i);
|
|
323
|
7867
|
|
|
7868 /* The suggested word may replace only part of the bad word, add
|
|
|
7869 * the not replaced part. */
|
|
|
7870 STRCPY(wcopy, stp->st_word);
|
|
|
7871 if (sug.su_badlen > stp->st_orglen)
|
|
|
7872 vim_strncpy(wcopy + STRLEN(wcopy),
|
|
|
7873 sug.su_badptr + stp->st_orglen,
|
|
|
7874 sug.su_badlen - stp->st_orglen);
|
|
474
|
7875 vim_snprintf((char *)IObuff, IOSIZE, "%2d", i + 1);
|
|
|
7876 #ifdef FEAT_RIGHTLEFT
|
|
|
7877 if (cmdmsg_rl)
|
|
|
7878 rl_mirror(IObuff);
|
|
|
7879 #endif
|
|
|
7880 msg_puts(IObuff);
|
|
|
7881
|
|
|
7882 vim_snprintf((char *)IObuff, IOSIZE, " \"%s\"", wcopy);
|
|
346
|
7883 msg_puts(IObuff);
|
|
|
7884
|
|
|
7885 /* The word may replace more than "su_badlen". */
|
|
|
7886 if (sug.su_badlen < stp->st_orglen)
|
|
|
7887 {
|
|
|
7888 vim_snprintf((char *)IObuff, IOSIZE, _(" < \"%.*s\""),
|
|
|
7889 stp->st_orglen, sug.su_badptr);
|
|
|
7890 msg_puts(IObuff);
|
|
|
7891 }
|
|
|
7892
|
|
324
|
7893 if (p_verbose > 0)
|
|
344
|
7894 {
|
|
346
|
7895 /* Add the score. */
|
|
351
|
7896 if (sps_flags & (SPS_DOUBLE | SPS_BEST))
|
|
474
|
7897 vim_snprintf((char *)IObuff, IOSIZE, " (%s%d - %d)",
|
|
344
|
7898 stp->st_salscore ? "s " : "",
|
|
|
7899 stp->st_score, stp->st_altscore);
|
|
|
7900 else
|
|
474
|
7901 vim_snprintf((char *)IObuff, IOSIZE, " (%d)",
|
|
346
|
7902 stp->st_score);
|
|
474
|
7903 #ifdef FEAT_RIGHTLEFT
|
|
|
7904 if (cmdmsg_rl)
|
|
|
7905 /* Mirror the numbers, but keep the leading space. */
|
|
|
7906 rl_mirror(IObuff + 1);
|
|
|
7907 #endif
|
|
346
|
7908 msg_advance(30);
|
|
|
7909 msg_puts(IObuff);
|
|
344
|
7910 }
|
|
323
|
7911 msg_putchar('\n');
|
|
|
7912 }
|
|
|
7913
|
|
474
|
7914 #ifdef FEAT_RIGHTLEFT
|
|
|
7915 cmdmsg_rl = FALSE;
|
|
|
7916 msg_col = 0;
|
|
|
7917 #endif
|
|
323
|
7918 /* Ask for choice. */
|
|
485
|
7919 selected = prompt_for_number(&mouse_used);
|
|
372
|
7920 if (mouse_used)
|
|
485
|
7921 selected -= lines_left;
|
|
|
7922 }
|
|
|
7923
|
|
|
7924 if (selected > 0 && selected <= sug.su_ga.ga_len && u_save_cursor() == OK)
|
|
|
7925 {
|
|
|
7926 /* Save the from and to text for :spellrepall. */
|
|
|
7927 stp = &SUG(sug.su_ga, selected - 1);
|
|
|
7928 repl_from = vim_strnsave(sug.su_badptr, stp->st_orglen);
|
|
|
7929 repl_to = vim_strsave(stp->st_word);
|
|
|
7930
|
|
|
7931 /* Replace the word. */
|
|
|
7932 p = alloc(STRLEN(line) - stp->st_orglen + STRLEN(stp->st_word) + 1);
|
|
|
7933 if (p != NULL)
|
|
|
7934 {
|
|
|
7935 c = sug.su_badptr - line;
|
|
|
7936 mch_memmove(p, line, c);
|
|
|
7937 STRCPY(p + c, stp->st_word);
|
|
|
7938 STRCAT(p, sug.su_badptr + stp->st_orglen);
|
|
|
7939 ml_replace(curwin->w_cursor.lnum, p, FALSE);
|
|
|
7940 curwin->w_cursor.col = c;
|
|
|
7941 changed_bytes(curwin->w_cursor.lnum, c);
|
|
|
7942
|
|
|
7943 /* For redo we use a change-word command. */
|
|
|
7944 ResetRedobuff();
|
|
|
7945 AppendToRedobuff((char_u *)"ciw");
|
|
|
7946 AppendToRedobuff(stp->st_word);
|
|
|
7947 AppendCharToRedobuff(ESC);
|
|
|
7948 }
|
|
|
7949 }
|
|
|
7950 else
|
|
|
7951 curwin->w_cursor = prev_cursor;
|
|
323
|
7952
|
|
344
|
7953 spell_find_cleanup(&sug);
|
|
|
7954 }
|
|
|
7955
|
|
|
7956 /*
|
|
475
|
7957 * Check if the word at line "lnum" column "col" is required to start with a
|
|
|
7958 * capital. This uses 'spellcapcheck' of the current buffer.
|
|
|
7959 */
|
|
|
7960 static int
|
|
|
7961 check_need_cap(lnum, col)
|
|
|
7962 linenr_T lnum;
|
|
|
7963 colnr_T col;
|
|
|
7964 {
|
|
|
7965 int need_cap = FALSE;
|
|
|
7966 char_u *line;
|
|
|
7967 char_u *line_copy = NULL;
|
|
|
7968 char_u *p;
|
|
|
7969 colnr_T endcol;
|
|
|
7970 regmatch_T regmatch;
|
|
|
7971
|
|
|
7972 if (curbuf->b_cap_prog == NULL)
|
|
|
7973 return FALSE;
|
|
|
7974
|
|
|
7975 line = ml_get_curline();
|
|
|
7976 endcol = 0;
|
|
|
7977 if ((int)(skipwhite(line) - line) >= (int)col)
|
|
|
7978 {
|
|
|
7979 /* At start of line, check if previous line is empty or sentence
|
|
|
7980 * ends there. */
|
|
|
7981 if (lnum == 1)
|
|
|
7982 need_cap = TRUE;
|
|
|
7983 else
|
|
|
7984 {
|
|
|
7985 line = ml_get(lnum - 1);
|
|
|
7986 if (*skipwhite(line) == NUL)
|
|
|
7987 need_cap = TRUE;
|
|
|
7988 else
|
|
|
7989 {
|
|
|
7990 /* Append a space in place of the line break. */
|
|
|
7991 line_copy = concat_str(line, (char_u *)" ");
|
|
|
7992 line = line_copy;
|
|
|
7993 endcol = STRLEN(line);
|
|
|
7994 }
|
|
|
7995 }
|
|
|
7996 }
|
|
|
7997 else
|
|
|
7998 endcol = col;
|
|
|
7999
|
|
|
8000 if (endcol > 0)
|
|
|
8001 {
|
|
|
8002 /* Check if sentence ends before the bad word. */
|
|
|
8003 regmatch.regprog = curbuf->b_cap_prog;
|
|
|
8004 regmatch.rm_ic = FALSE;
|
|
|
8005 p = line + endcol;
|
|
|
8006 for (;;)
|
|
|
8007 {
|
|
|
8008 mb_ptr_back(line, p);
|
|
|
8009 if (p == line || spell_iswordp_nmw(p))
|
|
|
8010 break;
|
|
|
8011 if (vim_regexec(®match, p, 0)
|
|
|
8012 && regmatch.endp[0] == line + endcol)
|
|
|
8013 {
|
|
|
8014 need_cap = TRUE;
|
|
|
8015 break;
|
|
|
8016 }
|
|
|
8017 }
|
|
|
8018 }
|
|
|
8019
|
|
|
8020 vim_free(line_copy);
|
|
|
8021
|
|
|
8022 return need_cap;
|
|
|
8023 }
|
|
|
8024
|
|
|
8025
|
|
|
8026 /*
|
|
372
|
8027 * ":spellrepall"
|
|
|
8028 */
|
|
|
8029 /*ARGSUSED*/
|
|
|
8030 void
|
|
|
8031 ex_spellrepall(eap)
|
|
|
8032 exarg_T *eap;
|
|
|
8033 {
|
|
|
8034 pos_T pos = curwin->w_cursor;
|
|
|
8035 char_u *frompat;
|
|
|
8036 int addlen;
|
|
|
8037 char_u *line;
|
|
|
8038 char_u *p;
|
|
|
8039 int save_ws = p_ws;
|
|
483
|
8040 linenr_T prev_lnum = 0;
|
|
372
|
8041
|
|
|
8042 if (repl_from == NULL || repl_to == NULL)
|
|
|
8043 {
|
|
|
8044 EMSG(_("E752: No previous spell replacement"));
|
|
|
8045 return;
|
|
|
8046 }
|
|
|
8047 addlen = STRLEN(repl_to) - STRLEN(repl_from);
|
|
|
8048
|
|
|
8049 frompat = alloc(STRLEN(repl_from) + 7);
|
|
|
8050 if (frompat == NULL)
|
|
|
8051 return;
|
|
|
8052 sprintf((char *)frompat, "\\V\\<%s\\>", repl_from);
|
|
|
8053 p_ws = FALSE;
|
|
|
8054
|
|
483
|
8055 sub_nsubs = 0;
|
|
|
8056 sub_nlines = 0;
|
|
372
|
8057 curwin->w_cursor.lnum = 0;
|
|
|
8058 while (!got_int)
|
|
|
8059 {
|
|
|
8060 if (do_search(NULL, '/', frompat, 1L, SEARCH_KEEP) == 0
|
|
|
8061 || u_save_cursor() == FAIL)
|
|
|
8062 break;
|
|
|
8063
|
|
|
8064 /* Only replace when the right word isn't there yet. This happens
|
|
|
8065 * when changing "etc" to "etc.". */
|
|
|
8066 line = ml_get_curline();
|
|
|
8067 if (addlen <= 0 || STRNCMP(line + curwin->w_cursor.col,
|
|
|
8068 repl_to, STRLEN(repl_to)) != 0)
|
|
|
8069 {
|
|
|
8070 p = alloc(STRLEN(line) + addlen + 1);
|
|
|
8071 if (p == NULL)
|
|
|
8072 break;
|
|
|
8073 mch_memmove(p, line, curwin->w_cursor.col);
|
|
|
8074 STRCPY(p + curwin->w_cursor.col, repl_to);
|
|
|
8075 STRCAT(p, line + curwin->w_cursor.col + STRLEN(repl_from));
|
|
|
8076 ml_replace(curwin->w_cursor.lnum, p, FALSE);
|
|
|
8077 changed_bytes(curwin->w_cursor.lnum, curwin->w_cursor.col);
|
|
483
|
8078
|
|
|
8079 if (curwin->w_cursor.lnum != prev_lnum)
|
|
|
8080 {
|
|
|
8081 ++sub_nlines;
|
|
|
8082 prev_lnum = curwin->w_cursor.lnum;
|
|
|
8083 }
|
|
|
8084 ++sub_nsubs;
|
|
372
|
8085 }
|
|
|
8086 curwin->w_cursor.col += STRLEN(repl_to);
|
|
|
8087 }
|
|
|
8088
|
|
|
8089 p_ws = save_ws;
|
|
|
8090 curwin->w_cursor = pos;
|
|
|
8091 vim_free(frompat);
|
|
|
8092
|
|
483
|
8093 if (sub_nsubs == 0)
|
|
372
|
8094 EMSG2(_("E753: Not found: %s"), repl_from);
|
|
483
|
8095 else
|
|
|
8096 do_sub_msg(FALSE);
|
|
372
|
8097 }
|
|
|
8098
|
|
|
8099 /*
|
|
344
|
8100 * Find spell suggestions for "word". Return them in the growarray "*gap" as
|
|
|
8101 * a list of allocated strings.
|
|
|
8102 */
|
|
|
8103 void
|
|
475
|
8104 spell_suggest_list(gap, word, maxcount, need_cap)
|
|
344
|
8105 garray_T *gap;
|
|
|
8106 char_u *word;
|
|
|
8107 int maxcount; /* maximum nr of suggestions */
|
|
475
|
8108 int need_cap; /* 'spellcapcheck' matched */
|
|
344
|
8109 {
|
|
|
8110 suginfo_T sug;
|
|
|
8111 int i;
|
|
|
8112 suggest_T *stp;
|
|
|
8113 char_u *wcopy;
|
|
|
8114
|
|
475
|
8115 spell_find_suggest(word, &sug, maxcount, FALSE, need_cap);
|
|
344
|
8116
|
|
|
8117 /* Make room in "gap". */
|
|
|
8118 ga_init2(gap, sizeof(char_u *), sug.su_ga.ga_len + 1);
|
|
|
8119 if (ga_grow(gap, sug.su_ga.ga_len) == FAIL)
|
|
|
8120 return;
|
|
|
8121
|
|
323
|
8122 for (i = 0; i < sug.su_ga.ga_len; ++i)
|
|
344
|
8123 {
|
|
|
8124 stp = &SUG(sug.su_ga, i);
|
|
|
8125
|
|
|
8126 /* The suggested word may replace only part of "word", add the not
|
|
|
8127 * replaced part. */
|
|
|
8128 wcopy = alloc(STRLEN(stp->st_word)
|
|
|
8129 + STRLEN(sug.su_badptr + stp->st_orglen) + 1);
|
|
|
8130 if (wcopy == NULL)
|
|
|
8131 break;
|
|
|
8132 STRCPY(wcopy, stp->st_word);
|
|
|
8133 STRCAT(wcopy, sug.su_badptr + stp->st_orglen);
|
|
|
8134 ((char_u **)gap->ga_data)[gap->ga_len++] = wcopy;
|
|
|
8135 }
|
|
|
8136
|
|
|
8137 spell_find_cleanup(&sug);
|
|
|
8138 }
|
|
|
8139
|
|
|
8140 /*
|
|
|
8141 * Find spell suggestions for the word at the start of "badptr".
|
|
|
8142 * Return the suggestions in "su->su_ga".
|
|
|
8143 * The maximum number of suggestions is "maxcount".
|
|
|
8144 * Note: does use info for the current window.
|
|
|
8145 * This is based on the mechanisms of Aspell, but completely reimplemented.
|
|
|
8146 */
|
|
|
8147 static void
|
|
388
|
8148 spell_find_suggest(badptr, su, maxcount, banbadword, need_cap)
|
|
344
|
8149 char_u *badptr;
|
|
|
8150 suginfo_T *su;
|
|
|
8151 int maxcount;
|
|
358
|
8152 int banbadword; /* don't include badword in suggestions */
|
|
388
|
8153 int need_cap; /* word should start with capital */
|
|
344
|
8154 {
|
|
385
|
8155 int attr = 0;
|
|
372
|
8156 char_u buf[MAXPATHL];
|
|
|
8157 char_u *p;
|
|
|
8158 int do_combine = FALSE;
|
|
|
8159 char_u *sps_copy;
|
|
|
8160 #ifdef FEAT_EVAL
|
|
|
8161 static int expr_busy = FALSE;
|
|
|
8162 #endif
|
|
385
|
8163 int c;
|
|
344
|
8164
|
|
|
8165 /*
|
|
|
8166 * Set the info in "*su".
|
|
|
8167 */
|
|
|
8168 vim_memset(su, 0, sizeof(suginfo_T));
|
|
|
8169 ga_init2(&su->su_ga, (int)sizeof(suggest_T), 10);
|
|
|
8170 ga_init2(&su->su_sga, (int)sizeof(suggest_T), 10);
|
|
355
|
8171 if (*badptr == NUL)
|
|
|
8172 return;
|
|
344
|
8173 hash_init(&su->su_banned);
|
|
|
8174
|
|
|
8175 su->su_badptr = badptr;
|
|
385
|
8176 su->su_badlen = spell_check(curwin, su->su_badptr, &attr, NULL);
|
|
344
|
8177 su->su_maxcount = maxcount;
|
|
372
|
8178 su->su_maxscore = SCORE_MAXINIT;
|
|
344
|
8179
|
|
|
8180 if (su->su_badlen >= MAXWLEN)
|
|
|
8181 su->su_badlen = MAXWLEN - 1; /* just in case */
|
|
|
8182 vim_strncpy(su->su_badword, su->su_badptr, su->su_badlen);
|
|
|
8183 (void)spell_casefold(su->su_badptr, su->su_badlen,
|
|
|
8184 su->su_fbadword, MAXWLEN);
|
|
346
|
8185 /* get caps flags for bad word */
|
|
474
|
8186 su->su_badflags = badword_captype(su->su_badptr,
|
|
|
8187 su->su_badptr + su->su_badlen);
|
|
388
|
8188 if (need_cap)
|
|
|
8189 su->su_badflags |= WF_ONECAP;
|
|
344
|
8190
|
|
385
|
8191 /* If the word is not capitalised and spell_check() doesn't consider the
|
|
|
8192 * word to be bad then it might need to be capitalised. Add a suggestion
|
|
|
8193 * for that. */
|
|
455
|
8194 c = PTR2CHAR(su->su_badptr);
|
|
385
|
8195 if (!SPELL_ISUPPER(c) && attr == 0)
|
|
|
8196 {
|
|
|
8197 make_case_word(su->su_badword, buf, WF_ONECAP);
|
|
|
8198 add_suggestion(su, &su->su_ga, buf, su->su_badlen, SCORE_ICASE,
|
|
|
8199 0, TRUE);
|
|
|
8200 }
|
|
|
8201
|
|
344
|
8202 /* Ban the bad word itself. It may appear in another region. */
|
|
358
|
8203 if (banbadword)
|
|
|
8204 add_banned(su, su->su_badword);
|
|
344
|
8205
|
|
372
|
8206 /* Make a copy of 'spellsuggest', because the expression may change it. */
|
|
|
8207 sps_copy = vim_strsave(p_sps);
|
|
|
8208 if (sps_copy == NULL)
|
|
|
8209 return;
|
|
|
8210
|
|
|
8211 /* Loop over the items in 'spellsuggest'. */
|
|
|
8212 for (p = sps_copy; *p != NUL; )
|
|
|
8213 {
|
|
|
8214 copy_option_part(&p, buf, MAXPATHL, ",");
|
|
|
8215
|
|
|
8216 if (STRNCMP(buf, "expr:", 5) == 0)
|
|
|
8217 {
|
|
|
8218 #ifdef FEAT_EVAL
|
|
375
|
8219 /* Evaluate an expression. Skip this when called recursively,
|
|
|
8220 * when using spellsuggest() in the expression. */
|
|
372
|
8221 if (!expr_busy)
|
|
|
8222 {
|
|
|
8223 expr_busy = TRUE;
|
|
|
8224 spell_suggest_expr(su, buf + 5);
|
|
|
8225 expr_busy = FALSE;
|
|
|
8226 }
|
|
|
8227 #endif
|
|
|
8228 }
|
|
|
8229 else if (STRNCMP(buf, "file:", 5) == 0)
|
|
|
8230 /* Use list of suggestions in a file. */
|
|
|
8231 spell_suggest_file(su, buf + 5);
|
|
|
8232 else
|
|
|
8233 {
|
|
|
8234 /* Use internal method. */
|
|
|
8235 spell_suggest_intern(su);
|
|
|
8236 if (sps_flags & SPS_DOUBLE)
|
|
|
8237 do_combine = TRUE;
|
|
|
8238 }
|
|
|
8239 }
|
|
|
8240
|
|
|
8241 vim_free(sps_copy);
|
|
|
8242
|
|
|
8243 if (do_combine)
|
|
|
8244 /* Combine the two list of suggestions. This must be done last,
|
|
|
8245 * because sorting changes the order again. */
|
|
|
8246 score_combine(su);
|
|
|
8247 }
|
|
|
8248
|
|
|
8249 #ifdef FEAT_EVAL
|
|
|
8250 /*
|
|
|
8251 * Find suggestions by evaluating expression "expr".
|
|
|
8252 */
|
|
|
8253 static void
|
|
|
8254 spell_suggest_expr(su, expr)
|
|
|
8255 suginfo_T *su;
|
|
|
8256 char_u *expr;
|
|
|
8257 {
|
|
|
8258 list_T *list;
|
|
|
8259 listitem_T *li;
|
|
|
8260 int score;
|
|
|
8261 char_u *p;
|
|
|
8262
|
|
|
8263 /* The work is split up in a few parts to avoid having to export
|
|
|
8264 * suginfo_T.
|
|
|
8265 * First evaluate the expression and get the resulting list. */
|
|
|
8266 list = eval_spell_expr(su->su_badword, expr);
|
|
|
8267 if (list != NULL)
|
|
|
8268 {
|
|
|
8269 /* Loop over the items in the list. */
|
|
|
8270 for (li = list->lv_first; li != NULL; li = li->li_next)
|
|
|
8271 if (li->li_tv.v_type == VAR_LIST)
|
|
|
8272 {
|
|
|
8273 /* Get the word and the score from the items. */
|
|
|
8274 score = get_spellword(li->li_tv.vval.v_list, &p);
|
|
|
8275 if (score >= 0)
|
|
|
8276 add_suggestion(su, &su->su_ga, p,
|
|
|
8277 su->su_badlen, score, 0, TRUE);
|
|
|
8278 }
|
|
|
8279 list_unref(list);
|
|
|
8280 }
|
|
|
8281
|
|
|
8282 /* Sort the suggestions and truncate at "maxcount". */
|
|
|
8283 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
8284 }
|
|
|
8285 #endif
|
|
|
8286
|
|
|
8287 /*
|
|
474
|
8288 * Find suggestions in file "fname". Used for "file:" in 'spellsuggest'.
|
|
372
|
8289 */
|
|
|
8290 static void
|
|
|
8291 spell_suggest_file(su, fname)
|
|
|
8292 suginfo_T *su;
|
|
|
8293 char_u *fname;
|
|
|
8294 {
|
|
|
8295 FILE *fd;
|
|
|
8296 char_u line[MAXWLEN * 2];
|
|
|
8297 char_u *p;
|
|
|
8298 int len;
|
|
|
8299 char_u cword[MAXWLEN];
|
|
|
8300
|
|
|
8301 /* Open the file. */
|
|
|
8302 fd = mch_fopen((char *)fname, "r");
|
|
|
8303 if (fd == NULL)
|
|
|
8304 {
|
|
|
8305 EMSG2(_(e_notopen), fname);
|
|
|
8306 return;
|
|
|
8307 }
|
|
|
8308
|
|
|
8309 /* Read it line by line. */
|
|
|
8310 while (!vim_fgets(line, MAXWLEN * 2, fd) && !got_int)
|
|
|
8311 {
|
|
|
8312 line_breakcheck();
|
|
|
8313
|
|
|
8314 p = vim_strchr(line, '/');
|
|
|
8315 if (p == NULL)
|
|
|
8316 continue; /* No Tab found, just skip the line. */
|
|
|
8317 *p++ = NUL;
|
|
|
8318 if (STRICMP(su->su_badword, line) == 0)
|
|
|
8319 {
|
|
|
8320 /* Match! Isolate the good word, until CR or NL. */
|
|
|
8321 for (len = 0; p[len] >= ' '; ++len)
|
|
|
8322 ;
|
|
|
8323 p[len] = NUL;
|
|
|
8324
|
|
|
8325 /* If the suggestion doesn't have specific case duplicate the case
|
|
|
8326 * of the bad word. */
|
|
|
8327 if (captype(p, NULL) == 0)
|
|
|
8328 {
|
|
|
8329 make_case_word(p, cword, su->su_badflags);
|
|
|
8330 p = cword;
|
|
|
8331 }
|
|
|
8332
|
|
|
8333 add_suggestion(su, &su->su_ga, p, su->su_badlen,
|
|
|
8334 SCORE_FILE, 0, TRUE);
|
|
|
8335 }
|
|
|
8336 }
|
|
|
8337
|
|
|
8338 fclose(fd);
|
|
|
8339
|
|
|
8340 /* Sort the suggestions and truncate at "maxcount". */
|
|
|
8341 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
8342 }
|
|
|
8343
|
|
|
8344 /*
|
|
|
8345 * Find suggestions for the internal method indicated by "sps_flags".
|
|
|
8346 */
|
|
|
8347 static void
|
|
|
8348 spell_suggest_intern(su)
|
|
|
8349 suginfo_T *su;
|
|
|
8350 {
|
|
344
|
8351 /*
|
|
346
|
8352 * 1. Try special cases, such as repeating a word: "the the" -> "the".
|
|
344
|
8353 *
|
|
|
8354 * Set a maximum score to limit the combination of operations that is
|
|
|
8355 * tried.
|
|
|
8356 */
|
|
346
|
8357 suggest_try_special(su);
|
|
|
8358
|
|
|
8359 /*
|
|
|
8360 * 2. Try inserting/deleting/swapping/changing a letter, use REP entries
|
|
|
8361 * from the .aff file and inserting a space (split the word).
|
|
|
8362 */
|
|
|
8363 suggest_try_change(su);
|
|
344
|
8364
|
|
|
8365 /* For the resulting top-scorers compute the sound-a-like score. */
|
|
|
8366 if (sps_flags & SPS_DOUBLE)
|
|
|
8367 score_comp_sal(su);
|
|
|
8368
|
|
|
8369 /*
|
|
346
|
8370 * 3. Try finding sound-a-like words.
|
|
344
|
8371 *
|
|
|
8372 * Only do this when we don't have a lot of suggestions yet, because it's
|
|
|
8373 * very slow and often doesn't find new suggestions.
|
|
|
8374 */
|
|
|
8375 if ((sps_flags & SPS_DOUBLE)
|
|
|
8376 || (!(sps_flags & SPS_FAST)
|
|
|
8377 && su->su_ga.ga_len < SUG_CLEAN_COUNT(su)))
|
|
|
8378 {
|
|
|
8379 /* Allow a higher score now. */
|
|
|
8380 su->su_maxscore = SCORE_MAXMAX;
|
|
346
|
8381 suggest_try_soundalike(su);
|
|
344
|
8382 }
|
|
|
8383
|
|
|
8384 /* When CTRL-C was hit while searching do show the results. */
|
|
|
8385 ui_breakcheck();
|
|
|
8386 if (got_int)
|
|
|
8387 {
|
|
|
8388 (void)vgetc();
|
|
|
8389 got_int = FALSE;
|
|
|
8390 }
|
|
|
8391
|
|
372
|
8392 if ((sps_flags & SPS_DOUBLE) == 0 && su->su_ga.ga_len != 0)
|
|
344
|
8393 {
|
|
|
8394 if (sps_flags & SPS_BEST)
|
|
|
8395 /* Adjust the word score for how it sounds like. */
|
|
|
8396 rescore_suggestions(su);
|
|
|
8397
|
|
|
8398 /* Sort the suggestions and truncate at "maxcount". */
|
|
372
|
8399 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
344
|
8400 }
|
|
|
8401 }
|
|
|
8402
|
|
|
8403 /*
|
|
|
8404 * Free the info put in "*su" by spell_find_suggest().
|
|
|
8405 */
|
|
|
8406 static void
|
|
|
8407 spell_find_cleanup(su)
|
|
|
8408 suginfo_T *su;
|
|
|
8409 {
|
|
|
8410 int i;
|
|
|
8411
|
|
|
8412 /* Free the suggestions. */
|
|
|
8413 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
8414 vim_free(SUG(su->su_ga, i).st_word);
|
|
|
8415 ga_clear(&su->su_ga);
|
|
|
8416 for (i = 0; i < su->su_sga.ga_len; ++i)
|
|
|
8417 vim_free(SUG(su->su_sga, i).st_word);
|
|
|
8418 ga_clear(&su->su_sga);
|
|
323
|
8419
|
|
|
8420 /* Free the banned words. */
|
|
344
|
8421 free_banned(su);
|
|
323
|
8422 }
|
|
|
8423
|
|
|
8424 /*
|
|
324
|
8425 * Make a copy of "word", with the first letter upper or lower cased, to
|
|
|
8426 * "wcopy[MAXWLEN]". "word" must not be empty.
|
|
|
8427 * The result is NUL terminated.
|
|
323
|
8428 */
|
|
|
8429 static void
|
|
324
|
8430 onecap_copy(word, wcopy, upper)
|
|
323
|
8431 char_u *word;
|
|
|
8432 char_u *wcopy;
|
|
|
8433 int upper; /* TRUE: first letter made upper case */
|
|
|
8434 {
|
|
|
8435 char_u *p;
|
|
|
8436 int c;
|
|
|
8437 int l;
|
|
|
8438
|
|
|
8439 p = word;
|
|
|
8440 #ifdef FEAT_MBYTE
|
|
|
8441 if (has_mbyte)
|
|
474
|
8442 c = mb_cptr2char_adv(&p);
|
|
323
|
8443 else
|
|
|
8444 #endif
|
|
|
8445 c = *p++;
|
|
|
8446 if (upper)
|
|
324
|
8447 c = SPELL_TOUPPER(c);
|
|
323
|
8448 else
|
|
324
|
8449 c = SPELL_TOFOLD(c);
|
|
323
|
8450 #ifdef FEAT_MBYTE
|
|
|
8451 if (has_mbyte)
|
|
|
8452 l = mb_char2bytes(c, wcopy);
|
|
|
8453 else
|
|
|
8454 #endif
|
|
|
8455 {
|
|
|
8456 l = 1;
|
|
|
8457 wcopy[0] = c;
|
|
|
8458 }
|
|
376
|
8459 vim_strncpy(wcopy + l, p, MAXWLEN - l - 1);
|
|
323
|
8460 }
|
|
|
8461
|
|
|
8462 /*
|
|
324
|
8463 * Make a copy of "word" with all the letters upper cased into
|
|
|
8464 * "wcopy[MAXWLEN]". The result is NUL terminated.
|
|
323
|
8465 */
|
|
|
8466 static void
|
|
|
8467 allcap_copy(word, wcopy)
|
|
|
8468 char_u *word;
|
|
|
8469 char_u *wcopy;
|
|
|
8470 {
|
|
|
8471 char_u *s;
|
|
|
8472 char_u *d;
|
|
|
8473 int c;
|
|
|
8474
|
|
|
8475 d = wcopy;
|
|
|
8476 for (s = word; *s != NUL; )
|
|
|
8477 {
|
|
|
8478 #ifdef FEAT_MBYTE
|
|
|
8479 if (has_mbyte)
|
|
474
|
8480 c = mb_cptr2char_adv(&s);
|
|
323
|
8481 else
|
|
|
8482 #endif
|
|
|
8483 c = *s++;
|
|
492
|
8484
|
|
|
8485 #ifdef FEAT_MBYTE
|
|
|
8486 /* We only change ß to SS when we are certain latin1 is used. It
|
|
|
8487 * would cause weird errors in other 8-bit encodings. */
|
|
|
8488 if (enc_latin1like && c == 0xdf)
|
|
|
8489 {
|
|
|
8490 c = 'S';
|
|
|
8491 if (d - wcopy >= MAXWLEN - 1)
|
|
|
8492 break;
|
|
|
8493 *d++ = c;
|
|
|
8494 }
|
|
|
8495 else
|
|
|
8496 #endif
|
|
|
8497 c = SPELL_TOUPPER(c);
|
|
323
|
8498
|
|
|
8499 #ifdef FEAT_MBYTE
|
|
|
8500 if (has_mbyte)
|
|
|
8501 {
|
|
|
8502 if (d - wcopy >= MAXWLEN - MB_MAXBYTES)
|
|
|
8503 break;
|
|
|
8504 d += mb_char2bytes(c, d);
|
|
|
8505 }
|
|
|
8506 else
|
|
|
8507 #endif
|
|
|
8508 {
|
|
|
8509 if (d - wcopy >= MAXWLEN - 1)
|
|
|
8510 break;
|
|
|
8511 *d++ = c;
|
|
|
8512 }
|
|
|
8513 }
|
|
|
8514 *d = NUL;
|
|
|
8515 }
|
|
|
8516
|
|
|
8517 /*
|
|
346
|
8518 * Try finding suggestions by recognizing specific situations.
|
|
|
8519 */
|
|
|
8520 static void
|
|
|
8521 suggest_try_special(su)
|
|
|
8522 suginfo_T *su;
|
|
|
8523 {
|
|
|
8524 char_u *p;
|
|
376
|
8525 size_t len;
|
|
346
|
8526 int c;
|
|
|
8527 char_u word[MAXWLEN];
|
|
|
8528
|
|
|
8529 /*
|
|
|
8530 * Recognize a word that is repeated: "the the".
|
|
|
8531 */
|
|
|
8532 p = skiptowhite(su->su_fbadword);
|
|
|
8533 len = p - su->su_fbadword;
|
|
|
8534 p = skipwhite(p);
|
|
|
8535 if (STRLEN(p) == len && STRNCMP(su->su_fbadword, p, len) == 0)
|
|
|
8536 {
|
|
|
8537 /* Include badflags: if the badword is onecap or allcap
|
|
|
8538 * use that for the goodword too: "The the" -> "The". */
|
|
|
8539 c = su->su_fbadword[len];
|
|
|
8540 su->su_fbadword[len] = NUL;
|
|
|
8541 make_case_word(su->su_fbadword, word, su->su_badflags);
|
|
|
8542 su->su_fbadword[len] = c;
|
|
351
|
8543 add_suggestion(su, &su->su_ga, word, su->su_badlen, SCORE_DEL, 0, TRUE);
|
|
346
|
8544 }
|
|
|
8545 }
|
|
|
8546
|
|
|
8547 /*
|
|
323
|
8548 * Try finding suggestions by adding/removing/swapping letters.
|
|
330
|
8549 *
|
|
|
8550 * This uses a state machine. At each node in the tree we try various
|
|
|
8551 * operations. When trying if an operation work "depth" is increased and the
|
|
|
8552 * stack[] is used to store info. This allows combinations, thus insert one
|
|
|
8553 * character, replace one and delete another. The number of changes is
|
|
|
8554 * limited by su->su_maxscore, checked in try_deeper().
|
|
485
|
8555 *
|
|
|
8556 * After implementing this I noticed an article by Kemal Oflazer that
|
|
|
8557 * describes something similar: "Error-tolerant Finite State Recognition with
|
|
|
8558 * Applications to Morphological Analysis and Spelling Correction" (1996).
|
|
|
8559 * The implementation in the article is simplified and requires a stack of
|
|
|
8560 * unknown depth. The implementation here only needs a stack depth of the
|
|
|
8561 * length of the word.
|
|
323
|
8562 */
|
|
|
8563 static void
|
|
346
|
8564 suggest_try_change(su)
|
|
323
|
8565 suginfo_T *su;
|
|
|
8566 {
|
|
|
8567 char_u fword[MAXWLEN]; /* copy of the bad word, case-folded */
|
|
|
8568 char_u tword[MAXWLEN]; /* good word collected so far */
|
|
|
8569 trystate_T stack[MAXWLEN];
|
|
483
|
8570 char_u preword[MAXWLEN * 3]; /* word found with proper case;
|
|
|
8571 * concatanation of prefix compound
|
|
|
8572 * words and split word. NUL terminated
|
|
|
8573 * when going deeper but not when coming
|
|
|
8574 * back. */
|
|
|
8575 char_u compflags[MAXWLEN]; /* compound flags, one for each word */
|
|
323
|
8576 trystate_T *sp;
|
|
|
8577 int newscore;
|
|
|
8578 langp_T *lp;
|
|
375
|
8579 char_u *byts, *fbyts, *pbyts;
|
|
|
8580 idx_T *idxs, *fidxs, *pidxs;
|
|
323
|
8581 int depth;
|
|
330
|
8582 int c, c2, c3;
|
|
390
|
8583 int n;
|
|
323
|
8584 int flags;
|
|
|
8585 garray_T *gap;
|
|
324
|
8586 idx_T arridx;
|
|
323
|
8587 int len;
|
|
|
8588 char_u *p;
|
|
|
8589 fromto_T *ftp;
|
|
330
|
8590 int fl = 0, tl;
|
|
346
|
8591 int repextra = 0; /* extra bytes in fword[] from REP item */
|
|
482
|
8592 slang_T *slang;
|
|
|
8593 int fword_ends;
|
|
323
|
8594
|
|
|
8595 /* We make a copy of the case-folded bad word, so that we can modify it
|
|
346
|
8596 * to find matches (esp. REP items). Append some more text, changing
|
|
|
8597 * chars after the bad word may help. */
|
|
323
|
8598 STRCPY(fword, su->su_fbadword);
|
|
346
|
8599 n = STRLEN(fword);
|
|
|
8600 p = su->su_badptr + su->su_badlen;
|
|
|
8601 (void)spell_casefold(p, STRLEN(p), fword + n, MAXWLEN - n);
|
|
323
|
8602
|
|
|
8603 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
8604 lp->lp_slang != NULL; ++lp)
|
|
|
8605 {
|
|
482
|
8606 slang = lp->lp_slang;
|
|
|
8607
|
|
323
|
8608 /*
|
|
|
8609 * Go through the whole case-fold tree, try changes at each node.
|
|
|
8610 * "tword[]" contains the word collected from nodes in the tree.
|
|
|
8611 * "fword[]" the word we are trying to match with (initially the bad
|
|
|
8612 * word).
|
|
|
8613 */
|
|
|
8614 depth = 0;
|
|
375
|
8615 sp = &stack[0];
|
|
483
|
8616 vim_memset(sp, 0, sizeof(trystate_T));
|
|
375
|
8617 sp->ts_curi = 1;
|
|
330
|
8618
|
|
|
8619 /*
|
|
375
|
8620 * When there are postponed prefixes we need to use these first. At
|
|
|
8621 * the end of the prefix we continue in the case-fold tree.
|
|
|
8622 */
|
|
482
|
8623 fbyts = slang->sl_fbyts;
|
|
|
8624 fidxs = slang->sl_fidxs;
|
|
|
8625 pbyts = slang->sl_pbyts;
|
|
|
8626 pidxs = slang->sl_pidxs;
|
|
375
|
8627 if (pbyts != NULL)
|
|
|
8628 {
|
|
|
8629 byts = pbyts;
|
|
|
8630 idxs = pidxs;
|
|
482
|
8631 sp->ts_prefixdepth = PFD_PREFIXTREE;
|
|
375
|
8632 sp->ts_state = STATE_NOPREFIX; /* try without prefix first */
|
|
|
8633 }
|
|
|
8634 else
|
|
|
8635 {
|
|
|
8636 byts = fbyts;
|
|
|
8637 idxs = fidxs;
|
|
482
|
8638 sp->ts_prefixdepth = PFD_NOPREFIX;
|
|
485
|
8639 sp->ts_state = STATE_START;
|
|
375
|
8640 }
|
|
|
8641
|
|
|
8642 /*
|
|
330
|
8643 * Loop to find all suggestions. At each round we either:
|
|
|
8644 * - For the current state try one operation, advance "ts_curi",
|
|
|
8645 * increase "depth".
|
|
|
8646 * - When a state is done go to the next, set "ts_state".
|
|
|
8647 * - When all states are tried decrease "depth".
|
|
|
8648 */
|
|
323
|
8649 while (depth >= 0 && !got_int)
|
|
|
8650 {
|
|
|
8651 sp = &stack[depth];
|
|
|
8652 switch (sp->ts_state)
|
|
|
8653 {
|
|
|
8654 case STATE_START:
|
|
375
|
8655 case STATE_NOPREFIX:
|
|
323
|
8656 /*
|
|
|
8657 * Start of node: Deal with NUL bytes, which means
|
|
|
8658 * tword[] may end here.
|
|
|
8659 */
|
|
|
8660 arridx = sp->ts_arridx; /* current node in the tree */
|
|
|
8661 len = byts[arridx]; /* bytes in this node */
|
|
|
8662 arridx += sp->ts_curi; /* index of current byte */
|
|
|
8663
|
|
482
|
8664 if (sp->ts_prefixdepth == PFD_PREFIXTREE)
|
|
375
|
8665 {
|
|
|
8666 /* Skip over the NUL bytes, we use them later. */
|
|
|
8667 for (n = 0; n < len && byts[arridx + n] == 0; ++n)
|
|
|
8668 ;
|
|
|
8669 sp->ts_curi += n;
|
|
|
8670
|
|
390
|
8671 /* Always past NUL bytes now. */
|
|
|
8672 n = (int)sp->ts_state;
|
|
|
8673 sp->ts_state = STATE_ENDNUL;
|
|
455
|
8674 sp->ts_save_badflags = su->su_badflags;
|
|
390
|
8675
|
|
375
|
8676 /* At end of a prefix or at start of prefixtree: check for
|
|
|
8677 * following word. */
|
|
390
|
8678 if (byts[arridx] == 0 || n == (int)STATE_NOPREFIX)
|
|
375
|
8679 {
|
|
455
|
8680 /* Set su->su_badflags to the caps type at this
|
|
|
8681 * position. Use the caps type until here for the
|
|
|
8682 * prefix itself. */
|
|
|
8683 #ifdef FEAT_MBYTE
|
|
|
8684 if (has_mbyte)
|
|
|
8685 n = nofold_len(fword, sp->ts_fidx, su->su_badptr);
|
|
|
8686 else
|
|
|
8687 #endif
|
|
|
8688 n = sp->ts_fidx;
|
|
474
|
8689 flags = badword_captype(su->su_badptr,
|
|
|
8690 su->su_badptr + n);
|
|
|
8691 su->su_badflags = badword_captype(su->su_badptr + n,
|
|
455
|
8692 su->su_badptr + su->su_badlen);
|
|
375
|
8693 ++depth;
|
|
|
8694 stack[depth] = stack[depth - 1];
|
|
|
8695 sp = &stack[depth];
|
|
|
8696 sp->ts_prefixdepth = depth - 1;
|
|
|
8697 byts = fbyts;
|
|
|
8698 idxs = fidxs;
|
|
|
8699 sp->ts_state = STATE_START;
|
|
|
8700 sp->ts_curi = 1; /* start just after length byte */
|
|
|
8701 sp->ts_arridx = 0;
|
|
|
8702
|
|
455
|
8703 /* Move the prefix to preword[] with the right case
|
|
|
8704 * and make find_keepcap_word() works. */
|
|
485
|
8705 tword[sp->ts_twordlen] = NUL;
|
|
|
8706 make_case_word(tword + sp->ts_splitoff,
|
|
|
8707 preword + sp->ts_prewordlen,
|
|
|
8708 flags);
|
|
|
8709 sp->ts_prewordlen = STRLEN(preword);
|
|
483
|
8710 sp->ts_splitoff = sp->ts_twordlen;
|
|
375
|
8711 }
|
|
|
8712 break;
|
|
|
8713 }
|
|
|
8714
|
|
346
|
8715 if (sp->ts_curi > len || byts[arridx] != 0)
|
|
323
|
8716 {
|
|
|
8717 /* Past bytes in node and/or past NUL bytes. */
|
|
|
8718 sp->ts_state = STATE_ENDNUL;
|
|
455
|
8719 sp->ts_save_badflags = su->su_badflags;
|
|
323
|
8720 break;
|
|
|
8721 }
|
|
|
8722
|
|
|
8723 /*
|
|
|
8724 * End of word in tree.
|
|
|
8725 */
|
|
|
8726 ++sp->ts_curi; /* eat one NUL byte */
|
|
|
8727
|
|
324
|
8728 flags = (int)idxs[arridx];
|
|
483
|
8729 fword_ends = (fword[sp->ts_fidx] == NUL
|
|
|
8730 || !spell_iswordp(fword + sp->ts_fidx, curbuf));
|
|
|
8731 tword[sp->ts_twordlen] = NUL;
|
|
323
|
8732
|
|
485
|
8733 if (sp->ts_prefixdepth <= PFD_NOTSPECIAL
|
|
|
8734 && (sp->ts_flags & TSF_PREFIXOK) == 0)
|
|
375
|
8735 {
|
|
|
8736 /* There was a prefix before the word. Check that the
|
|
|
8737 * prefix can be used with this word. */
|
|
|
8738 /* Count the length of the NULs in the prefix. If there
|
|
|
8739 * are none this must be the first try without a prefix.
|
|
|
8740 */
|
|
|
8741 n = stack[sp->ts_prefixdepth].ts_arridx;
|
|
|
8742 len = pbyts[n++];
|
|
|
8743 for (c = 0; c < len && pbyts[n + c] == 0; ++c)
|
|
|
8744 ;
|
|
|
8745 if (c > 0)
|
|
|
8746 {
|
|
390
|
8747 c = valid_word_prefix(c, n, flags,
|
|
483
|
8748 tword + sp->ts_splitoff, slang, FALSE);
|
|
375
|
8749 if (c == 0)
|
|
|
8750 break;
|
|
|
8751
|
|
|
8752 /* Use the WF_RARE flag for a rare prefix. */
|
|
|
8753 if (c & WF_RAREPFX)
|
|
|
8754 flags |= WF_RARE;
|
|
485
|
8755
|
|
|
8756 /* Tricky: when checking for both prefix and
|
|
|
8757 * compounding we run into the prefix flag first.
|
|
|
8758 * Remember that it's OK, so that we accept the prefix
|
|
|
8759 * when arriving at a compound flag. */
|
|
|
8760 sp->ts_flags |= TSF_PREFIXOK;
|
|
375
|
8761 }
|
|
|
8762 }
|
|
|
8763
|
|
485
|
8764 if (sp->ts_complen > sp->ts_compsplit)
|
|
|
8765 {
|
|
492
|
8766 if (slang->sl_nobreak)
|
|
|
8767 {
|
|
|
8768 /* There was a word before this word. When there was
|
|
|
8769 * no change in this word (it was correct) add the
|
|
|
8770 * first word as a suggestion. If this word was
|
|
|
8771 * corrected too, we need to check if a correct word
|
|
|
8772 * follows. */
|
|
|
8773 if (sp->ts_fidx - sp->ts_splitfidx
|
|
|
8774 == sp->ts_twordlen - sp->ts_splitoff
|
|
|
8775 && STRNCMP(fword + sp->ts_splitfidx,
|
|
|
8776 tword + sp->ts_splitoff,
|
|
|
8777 sp->ts_fidx - sp->ts_splitfidx) == 0)
|
|
|
8778 {
|
|
|
8779 preword[sp->ts_prewordlen] = NUL;
|
|
|
8780 add_suggestion(su, &su->su_ga, preword,
|
|
|
8781 sp->ts_splitfidx - repextra,
|
|
|
8782 sp->ts_score, 0, FALSE);
|
|
|
8783 break;
|
|
|
8784 }
|
|
|
8785 }
|
|
|
8786 else
|
|
|
8787 {
|
|
|
8788 /* There was a compound word before this word. If
|
|
|
8789 * this word does not support compounding then give up
|
|
|
8790 * (splitting is tried for the word without compound
|
|
|
8791 * flag). */
|
|
|
8792 if (((unsigned)flags >> 24) == 0
|
|
|
8793 || sp->ts_twordlen - sp->ts_splitoff
|
|
485
|
8794 < slang->sl_compminlen)
|
|
492
|
8795 break;
|
|
|
8796 compflags[sp->ts_complen] = ((unsigned)flags >> 24);
|
|
|
8797 compflags[sp->ts_complen + 1] = NUL;
|
|
|
8798 vim_strncpy(preword + sp->ts_prewordlen,
|
|
|
8799 tword + sp->ts_splitoff,
|
|
|
8800 sp->ts_twordlen - sp->ts_splitoff);
|
|
|
8801 p = preword;
|
|
|
8802 while (*skiptowhite(p) != NUL)
|
|
|
8803 p = skipwhite(skiptowhite(p));
|
|
|
8804 if (fword_ends && !can_compound(slang, p,
|
|
485
|
8805 compflags + sp->ts_compsplit))
|
|
492
|
8806 break;
|
|
|
8807
|
|
|
8808 /* Get pointer to last char of previous word. */
|
|
|
8809 p = preword + sp->ts_prewordlen;
|
|
|
8810 mb_ptr_back(preword, p);
|
|
|
8811 }
|
|
490
|
8812 }
|
|
|
8813 else
|
|
|
8814 p = NULL;
|
|
485
|
8815
|
|
323
|
8816 /*
|
|
|
8817 * Form the word with proper case in preword.
|
|
|
8818 * If there is a word from a previous split, append.
|
|
|
8819 */
|
|
|
8820 if (flags & WF_KEEPCAP)
|
|
|
8821 /* Must find the word in the keep-case tree. */
|
|
483
|
8822 find_keepcap_word(slang, tword + sp->ts_splitoff,
|
|
|
8823 preword + sp->ts_prewordlen);
|
|
323
|
8824 else
|
|
346
|
8825 {
|
|
323
|
8826 /* Include badflags: if the badword is onecap or allcap
|
|
346
|
8827 * use that for the goodword too. But if the badword is
|
|
|
8828 * allcap and it's only one char long use onecap. */
|
|
|
8829 c = su->su_badflags;
|
|
|
8830 if ((c & WF_ALLCAP)
|
|
|
8831 #ifdef FEAT_MBYTE
|
|
474
|
8832 && su->su_badlen == (*mb_ptr2len)(su->su_badptr)
|
|
346
|
8833 #else
|
|
|
8834 && su->su_badlen == 1
|
|
|
8835 #endif
|
|
|
8836 )
|
|
|
8837 c = WF_ONECAP;
|
|
490
|
8838 c |= flags;
|
|
|
8839
|
|
|
8840 /* When appending a compound word after a word character
|
|
|
8841 * don't use Onecap. */
|
|
|
8842 if (p != NULL && spell_iswordp_nmw(p))
|
|
|
8843 c &= ~WF_ONECAP;
|
|
483
|
8844 make_case_word(tword + sp->ts_splitoff,
|
|
490
|
8845 preword + sp->ts_prewordlen, c);
|
|
346
|
8846 }
|
|
323
|
8847
|
|
|
8848 /* Don't use a banned word. It may appear again as a good
|
|
|
8849 * word, thus remember it. */
|
|
|
8850 if (flags & WF_BANNED)
|
|
|
8851 {
|
|
483
|
8852 add_banned(su, preword + sp->ts_prewordlen);
|
|
323
|
8853 break;
|
|
|
8854 }
|
|
483
|
8855 if (was_banned(su, preword + sp->ts_prewordlen)
|
|
372
|
8856 || was_banned(su, preword))
|
|
323
|
8857 break;
|
|
|
8858
|
|
|
8859 newscore = 0;
|
|
|
8860 if ((flags & WF_REGION)
|
|
390
|
8861 && (((unsigned)flags >> 16) & lp->lp_region) == 0)
|
|
323
|
8862 newscore += SCORE_REGION;
|
|
|
8863 if (flags & WF_RARE)
|
|
|
8864 newscore += SCORE_RARE;
|
|
|
8865
|
|
346
|
8866 if (!spell_valid_case(su->su_badflags,
|
|
483
|
8867 captype(preword + sp->ts_prewordlen, NULL)))
|
|
323
|
8868 newscore += SCORE_ICASE;
|
|
|
8869
|
|
482
|
8870 if (fword_ends && sp->ts_fidx >= sp->ts_fidxtry)
|
|
323
|
8871 {
|
|
366
|
8872 /* The badword also ends: add suggestions. Give a penalty
|
|
|
8873 * when changing non-word char to word char, e.g., "thes,"
|
|
|
8874 * -> "these". */
|
|
|
8875 p = fword + sp->ts_fidx;
|
|
|
8876 #ifdef FEAT_MBYTE
|
|
|
8877 if (has_mbyte)
|
|
|
8878 mb_ptr_back(fword, p);
|
|
|
8879 else
|
|
|
8880 #endif
|
|
|
8881 --p;
|
|
376
|
8882 if (!spell_iswordp(p, curbuf))
|
|
366
|
8883 {
|
|
|
8884 p = preword + STRLEN(preword);
|
|
|
8885 #ifdef FEAT_MBYTE
|
|
|
8886 if (has_mbyte)
|
|
|
8887 mb_ptr_back(preword, p);
|
|
|
8888 else
|
|
|
8889 #endif
|
|
|
8890 --p;
|
|
376
|
8891 if (spell_iswordp(p, curbuf))
|
|
366
|
8892 newscore += SCORE_NONWORD;
|
|
|
8893 }
|
|
|
8894
|
|
344
|
8895 add_suggestion(su, &su->su_ga, preword,
|
|
346
|
8896 sp->ts_fidx - repextra,
|
|
351
|
8897 sp->ts_score + newscore, 0, FALSE);
|
|
323
|
8898 }
|
|
482
|
8899 else if ((sp->ts_fidx >= sp->ts_fidxtry || fword_ends)
|
|
330
|
8900 #ifdef FEAT_MBYTE
|
|
|
8901 /* Don't split halfway a character. */
|
|
|
8902 && (!has_mbyte || sp->ts_tcharlen == 0)
|
|
|
8903 #endif
|
|
|
8904 )
|
|
323
|
8905 {
|
|
482
|
8906 int try_compound;
|
|
|
8907
|
|
|
8908 /* Get here in two situations:
|
|
|
8909 * 1. The word in the tree ends but the badword continues:
|
|
|
8910 * If the word allows compounding try that. Otherwise
|
|
|
8911 * try a split by inserting a space. For both check
|
|
|
8912 * that a valid words starts at fword[sp->ts_fidx].
|
|
492
|
8913 * For NOBREAK do like compounding to be able to check
|
|
|
8914 * if the next word is valid.
|
|
482
|
8915 * 2. The badword does end, but it was due to a change
|
|
|
8916 * (e.g., a swap). No need to split, but do check that
|
|
|
8917 * the following word is valid.
|
|
|
8918 */
|
|
485
|
8919 try_compound = FALSE;
|
|
482
|
8920 if (!fword_ends
|
|
483
|
8921 && ((unsigned)flags >> 24) != 0
|
|
|
8922 && sp->ts_twordlen - sp->ts_splitoff
|
|
482
|
8923 >= slang->sl_compminlen
|
|
490
|
8924 && (slang->sl_compsylmax < MAXWLEN
|
|
|
8925 || sp->ts_complen + 1 - sp->ts_compsplit
|
|
|
8926 < slang->sl_compmax)
|
|
485
|
8927 && (vim_strchr(sp->ts_complen == sp->ts_compsplit
|
|
|
8928 ? slang->sl_compstartflags
|
|
|
8929 : slang->sl_compallflags,
|
|
483
|
8930 ((unsigned)flags >> 24)) != NULL))
|
|
|
8931 {
|
|
482
|
8932 try_compound = TRUE;
|
|
483
|
8933 compflags[sp->ts_complen] = ((unsigned)flags >> 24);
|
|
|
8934 compflags[sp->ts_complen + 1] = NUL;
|
|
|
8935 }
|
|
485
|
8936
|
|
492
|
8937 /* For NOBREAK we never try splitting, it won't make any
|
|
|
8938 * word valid. */
|
|
|
8939 if (slang->sl_nobreak)
|
|
|
8940 try_compound = TRUE;
|
|
|
8941
|
|
485
|
8942 /* If we could add a compound word, and it's also possible
|
|
|
8943 * to split at this point, do the split first and set
|
|
|
8944 * TSF_DIDSPLIT to avoid doing it again. */
|
|
492
|
8945 else if (!fword_ends
|
|
485
|
8946 && try_compound
|
|
|
8947 && (sp->ts_flags & TSF_DIDSPLIT) == 0)
|
|
482
|
8948 {
|
|
|
8949 try_compound = FALSE;
|
|
485
|
8950 sp->ts_flags |= TSF_DIDSPLIT;
|
|
|
8951 --sp->ts_curi; /* do the same NUL again */
|
|
|
8952 compflags[sp->ts_complen] = NUL;
|
|
|
8953 }
|
|
|
8954 else
|
|
|
8955 sp->ts_flags &= ~TSF_DIDSPLIT;
|
|
|
8956
|
|
|
8957 if (!try_compound && !fword_ends)
|
|
|
8958 {
|
|
|
8959 /* If we're going to split need to check that the
|
|
|
8960 * words so far are valid for compounding. */
|
|
490
|
8961 p = preword;
|
|
|
8962 while (*skiptowhite(p) != NUL)
|
|
|
8963 p = skipwhite(skiptowhite(p));
|
|
485
|
8964 if (sp->ts_complen > sp->ts_compsplit
|
|
490
|
8965 && !can_compound(slang, p,
|
|
485
|
8966 compflags + sp->ts_compsplit))
|
|
|
8967 break;
|
|
|
8968 newscore += SCORE_SPLIT;
|
|
482
|
8969 }
|
|
|
8970
|
|
|
8971 if (try_deeper(su, stack, depth, newscore))
|
|
323
|
8972 {
|
|
|
8973 /* Save things to be restored at STATE_SPLITUNDO. */
|
|
346
|
8974 sp->ts_save_badflags = su->su_badflags;
|
|
376
|
8975 sp->ts_state = STATE_SPLITUNDO;
|
|
|
8976
|
|
|
8977 ++depth;
|
|
|
8978 sp = &stack[depth];
|
|
323
|
8979
|
|
482
|
8980 /* Append a space to preword when splitting. */
|
|
|
8981 if (!try_compound && !fword_ends)
|
|
|
8982 STRCAT(preword, " ");
|
|
483
|
8983 sp->ts_prewordlen = STRLEN(preword);
|
|
|
8984 sp->ts_splitoff = sp->ts_twordlen;
|
|
492
|
8985 sp->ts_splitfidx = sp->ts_fidx;
|
|
376
|
8986
|
|
|
8987 /* If the badword has a non-word character at this
|
|
|
8988 * position skip it. That means replacing the
|
|
482
|
8989 * non-word character with a space. Always skip a
|
|
|
8990 * character when the word ends. */
|
|
|
8991 if ((!try_compound
|
|
|
8992 && !spell_iswordp_nmw(fword + sp->ts_fidx))
|
|
|
8993 || fword_ends)
|
|
376
|
8994 {
|
|
482
|
8995 int l;
|
|
|
8996
|
|
376
|
8997 #ifdef FEAT_MBYTE
|
|
|
8998 if (has_mbyte)
|
|
482
|
8999 l = MB_BYTE2LEN(fword[sp->ts_fidx]);
|
|
376
|
9000 else
|
|
|
9001 #endif
|
|
482
|
9002 l = 1;
|
|
|
9003 if (fword_ends)
|
|
|
9004 {
|
|
|
9005 /* Copy the skipped character to preword. */
|
|
483
|
9006 mch_memmove(preword + sp->ts_prewordlen,
|
|
482
|
9007 fword + sp->ts_fidx, l);
|
|
483
|
9008 sp->ts_prewordlen += l;
|
|
|
9009 preword[sp->ts_prewordlen] = NUL;
|
|
482
|
9010 }
|
|
|
9011 else
|
|
|
9012 sp->ts_score -= SCORE_SPLIT - SCORE_SUBST;
|
|
|
9013 sp->ts_fidx += l;
|
|
376
|
9014 }
|
|
455
|
9015
|
|
485
|
9016 /* When compounding include compound flag in
|
|
|
9017 * compflags[] (already set above). When splitting we
|
|
|
9018 * may start compounding over again. */
|
|
482
|
9019 if (try_compound)
|
|
483
|
9020 ++sp->ts_complen;
|
|
482
|
9021 else
|
|
485
|
9022 sp->ts_compsplit = sp->ts_complen;
|
|
|
9023 sp->ts_prefixdepth = PFD_NOPREFIX;
|
|
482
|
9024
|
|
455
|
9025 /* set su->su_badflags to the caps type at this
|
|
|
9026 * position */
|
|
324
|
9027 #ifdef FEAT_MBYTE
|
|
|
9028 if (has_mbyte)
|
|
455
|
9029 n = nofold_len(fword, sp->ts_fidx, su->su_badptr);
|
|
324
|
9030 else
|
|
|
9031 #endif
|
|
455
|
9032 n = sp->ts_fidx;
|
|
474
|
9033 su->su_badflags = badword_captype(su->su_badptr + n,
|
|
455
|
9034 su->su_badptr + su->su_badlen);
|
|
323
|
9035
|
|
|
9036 /* Restart at top of the tree. */
|
|
376
|
9037 sp->ts_arridx = 0;
|
|
485
|
9038
|
|
|
9039 /* If there are postponed prefixes, try these too. */
|
|
|
9040 if (pbyts != NULL)
|
|
|
9041 {
|
|
|
9042 byts = pbyts;
|
|
|
9043 idxs = pidxs;
|
|
|
9044 sp->ts_prefixdepth = PFD_PREFIXTREE;
|
|
|
9045 sp->ts_state = STATE_NOPREFIX;
|
|
|
9046 }
|
|
323
|
9047 }
|
|
|
9048 }
|
|
|
9049 break;
|
|
|
9050
|
|
|
9051 case STATE_SPLITUNDO:
|
|
482
|
9052 /* Undo the changes done for word split or compound word. */
|
|
346
|
9053 su->su_badflags = sp->ts_save_badflags;
|
|
323
|
9054
|
|
|
9055 /* Continue looking for NUL bytes. */
|
|
|
9056 sp->ts_state = STATE_START;
|
|
485
|
9057
|
|
|
9058 /* In case we went into the prefix tree. */
|
|
|
9059 byts = fbyts;
|
|
|
9060 idxs = fidxs;
|
|
323
|
9061 break;
|
|
|
9062
|
|
|
9063 case STATE_ENDNUL:
|
|
|
9064 /* Past the NUL bytes in the node. */
|
|
455
|
9065 su->su_badflags = sp->ts_save_badflags;
|
|
346
|
9066 if (fword[sp->ts_fidx] == NUL)
|
|
323
|
9067 {
|
|
|
9068 /* The badword ends, can't use the bytes in this node. */
|
|
|
9069 sp->ts_state = STATE_DEL;
|
|
|
9070 break;
|
|
|
9071 }
|
|
|
9072 sp->ts_state = STATE_PLAIN;
|
|
|
9073 /*FALLTHROUGH*/
|
|
|
9074
|
|
|
9075 case STATE_PLAIN:
|
|
|
9076 /*
|
|
|
9077 * Go over all possible bytes at this node, add each to
|
|
|
9078 * tword[] and use child node. "ts_curi" is the index.
|
|
|
9079 */
|
|
|
9080 arridx = sp->ts_arridx;
|
|
|
9081 if (sp->ts_curi > byts[arridx])
|
|
|
9082 {
|
|
|
9083 /* Done all bytes at this node, do next state. When still
|
|
|
9084 * at already changed bytes skip the other tricks. */
|
|
|
9085 if (sp->ts_fidx >= sp->ts_fidxtry)
|
|
|
9086 sp->ts_state = STATE_DEL;
|
|
|
9087 else
|
|
|
9088 sp->ts_state = STATE_FINAL;
|
|
|
9089 }
|
|
|
9090 else
|
|
|
9091 {
|
|
|
9092 arridx += sp->ts_curi++;
|
|
|
9093 c = byts[arridx];
|
|
|
9094
|
|
|
9095 /* Normal byte, go one level deeper. If it's not equal to
|
|
|
9096 * the byte in the bad word adjust the score. But don't
|
|
|
9097 * even try when the byte was already changed. */
|
|
330
|
9098 if (c == fword[sp->ts_fidx]
|
|
|
9099 #ifdef FEAT_MBYTE
|
|
|
9100 || (sp->ts_tcharlen > 0
|
|
|
9101 && sp->ts_isdiff != DIFF_NONE)
|
|
|
9102 #endif
|
|
|
9103 )
|
|
323
|
9104 newscore = 0;
|
|
|
9105 else
|
|
|
9106 newscore = SCORE_SUBST;
|
|
|
9107 if ((newscore == 0 || sp->ts_fidx >= sp->ts_fidxtry)
|
|
|
9108 && try_deeper(su, stack, depth, newscore))
|
|
|
9109 {
|
|
|
9110 ++depth;
|
|
330
|
9111 sp = &stack[depth];
|
|
|
9112 ++sp->ts_fidx;
|
|
|
9113 tword[sp->ts_twordlen++] = c;
|
|
|
9114 sp->ts_arridx = idxs[arridx];
|
|
|
9115 #ifdef FEAT_MBYTE
|
|
|
9116 if (newscore == SCORE_SUBST)
|
|
|
9117 sp->ts_isdiff = DIFF_YES;
|
|
|
9118 if (has_mbyte)
|
|
|
9119 {
|
|
|
9120 /* Multi-byte characters are a bit complicated to
|
|
|
9121 * handle: They differ when any of the bytes
|
|
|
9122 * differ and then their length may also differ. */
|
|
|
9123 if (sp->ts_tcharlen == 0)
|
|
|
9124 {
|
|
|
9125 /* First byte. */
|
|
|
9126 sp->ts_tcharidx = 0;
|
|
|
9127 sp->ts_tcharlen = MB_BYTE2LEN(c);
|
|
|
9128 sp->ts_fcharstart = sp->ts_fidx - 1;
|
|
|
9129 sp->ts_isdiff = (newscore != 0)
|
|
|
9130 ? DIFF_YES : DIFF_NONE;
|
|
|
9131 }
|
|
|
9132 else if (sp->ts_isdiff == DIFF_INSERT)
|
|
|
9133 /* When inserting trail bytes don't advance in
|
|
|
9134 * the bad word. */
|
|
|
9135 --sp->ts_fidx;
|
|
|
9136 if (++sp->ts_tcharidx == sp->ts_tcharlen)
|
|
|
9137 {
|
|
|
9138 /* Last byte of character. */
|
|
|
9139 if (sp->ts_isdiff == DIFF_YES)
|
|
|
9140 {
|
|
|
9141 /* Correct ts_fidx for the byte length of
|
|
|
9142 * the character (we didn't check that
|
|
|
9143 * before). */
|
|
|
9144 sp->ts_fidx = sp->ts_fcharstart
|
|
|
9145 + MB_BYTE2LEN(
|
|
|
9146 fword[sp->ts_fcharstart]);
|
|
|
9147
|
|
479
|
9148 /* For changing a composing character
|
|
|
9149 * adjust the score from SCORE_SUBST to
|
|
|
9150 * SCORE_SUBCOMP. */
|
|
|
9151 if (enc_utf8
|
|
|
9152 && utf_iscomposing(
|
|
|
9153 mb_ptr2char(tword
|
|
|
9154 + sp->ts_twordlen
|
|
|
9155 - sp->ts_tcharlen))
|
|
|
9156 && utf_iscomposing(
|
|
|
9157 mb_ptr2char(fword
|
|
|
9158 + sp->ts_fcharstart)))
|
|
|
9159 sp->ts_score -=
|
|
|
9160 SCORE_SUBST - SCORE_SUBCOMP;
|
|
|
9161
|
|
330
|
9162 /* For a similar character adjust score
|
|
|
9163 * from SCORE_SUBST to SCORE_SIMILAR. */
|
|
482
|
9164 else if (slang->sl_has_map
|
|
|
9165 && similar_chars(slang,
|
|
330
|
9166 mb_ptr2char(tword
|
|
|
9167 + sp->ts_twordlen
|
|
|
9168 - sp->ts_tcharlen),
|
|
|
9169 mb_ptr2char(fword
|
|
|
9170 + sp->ts_fcharstart)))
|
|
|
9171 sp->ts_score -=
|
|
|
9172 SCORE_SUBST - SCORE_SIMILAR;
|
|
|
9173 }
|
|
358
|
9174 else if (sp->ts_isdiff == DIFF_INSERT
|
|
|
9175 && sp->ts_twordlen > sp->ts_tcharlen)
|
|
|
9176 {
|
|
|
9177 p = tword + sp->ts_twordlen
|
|
|
9178 - sp->ts_tcharlen;
|
|
|
9179 c = mb_ptr2char(p);
|
|
475
|
9180 if (enc_utf8 && utf_iscomposing(c))
|
|
|
9181 {
|
|
|
9182 /* Inserting a composing char doesn't
|
|
|
9183 * count that much. */
|
|
358
|
9184 sp->ts_score -= SCORE_INS
|
|
475
|
9185 - SCORE_INSCOMP;
|
|
|
9186 }
|
|
|
9187 else
|
|
|
9188 {
|
|
|
9189 /* If the previous character was the
|
|
|
9190 * same, thus doubling a character,
|
|
|
9191 * give a bonus to the score. */
|
|
|
9192 mb_ptr_back(tword, p);
|
|
|
9193 if (c == mb_ptr2char(p))
|
|
|
9194 sp->ts_score -= SCORE_INS
|
|
358
|
9195 - SCORE_INSDUP;
|
|
475
|
9196 }
|
|
358
|
9197 }
|
|
330
|
9198
|
|
|
9199 /* Starting a new char, reset the length. */
|
|
|
9200 sp->ts_tcharlen = 0;
|
|
|
9201 }
|
|
|
9202 }
|
|
|
9203 else
|
|
|
9204 #endif
|
|
|
9205 {
|
|
|
9206 /* If we found a similar char adjust the score.
|
|
|
9207 * We do this after calling try_deeper() because
|
|
|
9208 * it's slow. */
|
|
|
9209 if (newscore != 0
|
|
482
|
9210 && slang->sl_has_map
|
|
|
9211 && similar_chars(slang,
|
|
330
|
9212 c, fword[sp->ts_fidx - 1]))
|
|
|
9213 sp->ts_score -= SCORE_SUBST - SCORE_SIMILAR;
|
|
|
9214 }
|
|
323
|
9215 }
|
|
|
9216 }
|
|
|
9217 break;
|
|
|
9218
|
|
|
9219 case STATE_DEL:
|
|
330
|
9220 #ifdef FEAT_MBYTE
|
|
|
9221 /* When past the first byte of a multi-byte char don't try
|
|
|
9222 * delete/insert/swap a character. */
|
|
|
9223 if (has_mbyte && sp->ts_tcharlen > 0)
|
|
|
9224 {
|
|
|
9225 sp->ts_state = STATE_FINAL;
|
|
|
9226 break;
|
|
|
9227 }
|
|
|
9228 #endif
|
|
|
9229 /*
|
|
|
9230 * Try skipping one character in the bad word (delete it).
|
|
|
9231 */
|
|
323
|
9232 sp->ts_state = STATE_INS;
|
|
|
9233 sp->ts_curi = 1;
|
|
|
9234 if (fword[sp->ts_fidx] != NUL
|
|
|
9235 && try_deeper(su, stack, depth, SCORE_DEL))
|
|
|
9236 {
|
|
|
9237 ++depth;
|
|
358
|
9238
|
|
|
9239 /* Advance over the character in fword[]. Give a bonus to
|
|
|
9240 * the score if the same character is following "nn" ->
|
|
|
9241 * "n". */
|
|
330
|
9242 #ifdef FEAT_MBYTE
|
|
|
9243 if (has_mbyte)
|
|
358
|
9244 {
|
|
|
9245 c = mb_ptr2char(fword + sp->ts_fidx);
|
|
330
|
9246 stack[depth].ts_fidx += MB_BYTE2LEN(fword[sp->ts_fidx]);
|
|
479
|
9247 if (enc_utf8 && utf_iscomposing(c))
|
|
|
9248 stack[depth].ts_score -= SCORE_DEL - SCORE_DELCOMP;
|
|
|
9249 else if (c == mb_ptr2char(fword + stack[depth].ts_fidx))
|
|
358
|
9250 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP;
|
|
|
9251 }
|
|
330
|
9252 else
|
|
|
9253 #endif
|
|
358
|
9254 {
|
|
330
|
9255 ++stack[depth].ts_fidx;
|
|
358
|
9256 if (fword[sp->ts_fidx] == fword[sp->ts_fidx + 1])
|
|
|
9257 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP;
|
|
|
9258 }
|
|
323
|
9259 break;
|
|
|
9260 }
|
|
|
9261 /*FALLTHROUGH*/
|
|
|
9262
|
|
|
9263 case STATE_INS:
|
|
330
|
9264 /* Insert one byte. Do this for each possible byte at this
|
|
323
|
9265 * node. */
|
|
|
9266 n = sp->ts_arridx;
|
|
|
9267 if (sp->ts_curi > byts[n])
|
|
|
9268 {
|
|
|
9269 /* Done all bytes at this node, do next state. */
|
|
|
9270 sp->ts_state = STATE_SWAP;
|
|
|
9271 }
|
|
|
9272 else
|
|
|
9273 {
|
|
330
|
9274 /* Do one more byte at this node. Skip NUL bytes. */
|
|
323
|
9275 n += sp->ts_curi++;
|
|
|
9276 c = byts[n];
|
|
|
9277 if (c != 0 && try_deeper(su, stack, depth, SCORE_INS))
|
|
|
9278 {
|
|
|
9279 ++depth;
|
|
330
|
9280 sp = &stack[depth];
|
|
|
9281 tword[sp->ts_twordlen++] = c;
|
|
|
9282 sp->ts_arridx = idxs[n];
|
|
|
9283 #ifdef FEAT_MBYTE
|
|
|
9284 if (has_mbyte)
|
|
|
9285 {
|
|
|
9286 fl = MB_BYTE2LEN(c);
|
|
|
9287 if (fl > 1)
|
|
|
9288 {
|
|
|
9289 /* There are following bytes for the same
|
|
|
9290 * character. We must find all bytes before
|
|
|
9291 * trying delete/insert/swap/etc. */
|
|
|
9292 sp->ts_tcharlen = fl;
|
|
|
9293 sp->ts_tcharidx = 1;
|
|
|
9294 sp->ts_isdiff = DIFF_INSERT;
|
|
|
9295 }
|
|
|
9296 }
|
|
358
|
9297 else
|
|
|
9298 fl = 1;
|
|
|
9299 if (fl == 1)
|
|
330
|
9300 #endif
|
|
358
|
9301 {
|
|
|
9302 /* If the previous character was the same, thus
|
|
|
9303 * doubling a character, give a bonus to the
|
|
|
9304 * score. */
|
|
|
9305 if (sp->ts_twordlen >= 2
|
|
|
9306 && tword[sp->ts_twordlen - 2] == c)
|
|
|
9307 sp->ts_score -= SCORE_INS - SCORE_INSDUP;
|
|
|
9308 }
|
|
323
|
9309 }
|
|
|
9310 }
|
|
|
9311 break;
|
|
|
9312
|
|
|
9313 case STATE_SWAP:
|
|
330
|
9314 /*
|
|
|
9315 * Swap two bytes in the bad word: "12" -> "21".
|
|
|
9316 * We change "fword" here, it's changed back afterwards.
|
|
|
9317 */
|
|
|
9318 p = fword + sp->ts_fidx;
|
|
|
9319 c = *p;
|
|
|
9320 if (c == NUL)
|
|
|
9321 {
|
|
|
9322 /* End of word, can't swap or replace. */
|
|
|
9323 sp->ts_state = STATE_FINAL;
|
|
|
9324 break;
|
|
|
9325 }
|
|
|
9326 #ifdef FEAT_MBYTE
|
|
|
9327 if (has_mbyte)
|
|
|
9328 {
|
|
474
|
9329 n = mb_cptr2len(p);
|
|
330
|
9330 c = mb_ptr2char(p);
|
|
|
9331 c2 = mb_ptr2char(p + n);
|
|
|
9332 }
|
|
|
9333 else
|
|
|
9334 #endif
|
|
|
9335 c2 = p[1];
|
|
|
9336 if (c == c2)
|
|
323
|
9337 {
|
|
330
|
9338 /* Characters are identical, swap won't do anything. */
|
|
|
9339 sp->ts_state = STATE_SWAP3;
|
|
|
9340 break;
|
|
|
9341 }
|
|
|
9342 if (c2 != NUL && try_deeper(su, stack, depth, SCORE_SWAP))
|
|
|
9343 {
|
|
|
9344 sp->ts_state = STATE_UNSWAP;
|
|
323
|
9345 ++depth;
|
|
330
|
9346 #ifdef FEAT_MBYTE
|
|
|
9347 if (has_mbyte)
|
|
|
9348 {
|
|
|
9349 fl = mb_char2len(c2);
|
|
|
9350 mch_memmove(p, p + n, fl);
|
|
|
9351 mb_char2bytes(c, p + fl);
|
|
|
9352 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl;
|
|
|
9353 }
|
|
|
9354 else
|
|
|
9355 #endif
|
|
|
9356 {
|
|
|
9357 p[0] = c2;
|
|
|
9358 p[1] = c;
|
|
|
9359 stack[depth].ts_fidxtry = sp->ts_fidx + 2;
|
|
|
9360 }
|
|
323
|
9361 }
|
|
|
9362 else
|
|
|
9363 /* If this swap doesn't work then SWAP3 won't either. */
|
|
|
9364 sp->ts_state = STATE_REP_INI;
|
|
|
9365 break;
|
|
|
9366
|
|
330
|
9367 case STATE_UNSWAP:
|
|
|
9368 /* Undo the STATE_SWAP swap: "21" -> "12". */
|
|
|
9369 p = fword + sp->ts_fidx;
|
|
|
9370 #ifdef FEAT_MBYTE
|
|
|
9371 if (has_mbyte)
|
|
|
9372 {
|
|
|
9373 n = MB_BYTE2LEN(*p);
|
|
|
9374 c = mb_ptr2char(p + n);
|
|
|
9375 mch_memmove(p + MB_BYTE2LEN(p[n]), p, n);
|
|
|
9376 mb_char2bytes(c, p);
|
|
|
9377 }
|
|
|
9378 else
|
|
|
9379 #endif
|
|
|
9380 {
|
|
|
9381 c = *p;
|
|
|
9382 *p = p[1];
|
|
|
9383 p[1] = c;
|
|
|
9384 }
|
|
|
9385 /*FALLTHROUGH*/
|
|
|
9386
|
|
|
9387 case STATE_SWAP3:
|
|
323
|
9388 /* Swap two bytes, skipping one: "123" -> "321". We change
|
|
330
|
9389 * "fword" here, it's changed back afterwards. */
|
|
|
9390 p = fword + sp->ts_fidx;
|
|
|
9391 #ifdef FEAT_MBYTE
|
|
|
9392 if (has_mbyte)
|
|
|
9393 {
|
|
474
|
9394 n = mb_cptr2len(p);
|
|
330
|
9395 c = mb_ptr2char(p);
|
|
474
|
9396 fl = mb_cptr2len(p + n);
|
|
330
|
9397 c2 = mb_ptr2char(p + n);
|
|
|
9398 c3 = mb_ptr2char(p + n + fl);
|
|
|
9399 }
|
|
|
9400 else
|
|
|
9401 #endif
|
|
323
|
9402 {
|
|
330
|
9403 c = *p;
|
|
|
9404 c2 = p[1];
|
|
|
9405 c3 = p[2];
|
|
|
9406 }
|
|
|
9407
|
|
|
9408 /* When characters are identical: "121" then SWAP3 result is
|
|
|
9409 * identical, ROT3L result is same as SWAP: "211", ROT3L
|
|
|
9410 * result is same as SWAP on next char: "112". Thus skip all
|
|
|
9411 * swapping. Also skip when c3 is NUL. */
|
|
|
9412 if (c == c3 || c3 == NUL)
|
|
|
9413 {
|
|
|
9414 sp->ts_state = STATE_REP_INI;
|
|
|
9415 break;
|
|
|
9416 }
|
|
|
9417 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
9418 {
|
|
|
9419 sp->ts_state = STATE_UNSWAP3;
|
|
323
|
9420 ++depth;
|
|
330
|
9421 #ifdef FEAT_MBYTE
|
|
|
9422 if (has_mbyte)
|
|
|
9423 {
|
|
|
9424 tl = mb_char2len(c3);
|
|
|
9425 mch_memmove(p, p + n + fl, tl);
|
|
|
9426 mb_char2bytes(c2, p + tl);
|
|
|
9427 mb_char2bytes(c, p + fl + tl);
|
|
|
9428 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl + tl;
|
|
|
9429 }
|
|
|
9430 else
|
|
|
9431 #endif
|
|
|
9432 {
|
|
|
9433 p[0] = p[2];
|
|
|
9434 p[2] = c;
|
|
|
9435 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
9436 }
|
|
323
|
9437 }
|
|
|
9438 else
|
|
|
9439 sp->ts_state = STATE_REP_INI;
|
|
|
9440 break;
|
|
|
9441
|
|
330
|
9442 case STATE_UNSWAP3:
|
|
|
9443 /* Undo STATE_SWAP3: "321" -> "123" */
|
|
|
9444 p = fword + sp->ts_fidx;
|
|
|
9445 #ifdef FEAT_MBYTE
|
|
|
9446 if (has_mbyte)
|
|
|
9447 {
|
|
|
9448 n = MB_BYTE2LEN(*p);
|
|
|
9449 c2 = mb_ptr2char(p + n);
|
|
|
9450 fl = MB_BYTE2LEN(p[n]);
|
|
|
9451 c = mb_ptr2char(p + n + fl);
|
|
|
9452 tl = MB_BYTE2LEN(p[n + fl]);
|
|
|
9453 mch_memmove(p + fl + tl, p, n);
|
|
|
9454 mb_char2bytes(c, p);
|
|
|
9455 mb_char2bytes(c2, p + tl);
|
|
|
9456 }
|
|
|
9457 else
|
|
|
9458 #endif
|
|
|
9459 {
|
|
|
9460 c = *p;
|
|
|
9461 *p = p[2];
|
|
|
9462 p[2] = c;
|
|
|
9463 }
|
|
346
|
9464
|
|
330
|
9465 /* Rotate three characters left: "123" -> "231". We change
|
|
|
9466 * "fword" here, it's changed back afterwards. */
|
|
323
|
9467 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
9468 {
|
|
330
|
9469 sp->ts_state = STATE_UNROT3L;
|
|
323
|
9470 ++depth;
|
|
330
|
9471 p = fword + sp->ts_fidx;
|
|
|
9472 #ifdef FEAT_MBYTE
|
|
|
9473 if (has_mbyte)
|
|
|
9474 {
|
|
474
|
9475 n = mb_cptr2len(p);
|
|
330
|
9476 c = mb_ptr2char(p);
|
|
474
|
9477 fl = mb_cptr2len(p + n);
|
|
|
9478 fl += mb_cptr2len(p + n + fl);
|
|
330
|
9479 mch_memmove(p, p + n, fl);
|
|
|
9480 mb_char2bytes(c, p + fl);
|
|
|
9481 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl;
|
|
|
9482 }
|
|
|
9483 else
|
|
|
9484 #endif
|
|
|
9485 {
|
|
|
9486 c = *p;
|
|
|
9487 *p = p[1];
|
|
|
9488 p[1] = p[2];
|
|
|
9489 p[2] = c;
|
|
|
9490 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
9491 }
|
|
323
|
9492 }
|
|
|
9493 else
|
|
|
9494 sp->ts_state = STATE_REP_INI;
|
|
|
9495 break;
|
|
|
9496
|
|
330
|
9497 case STATE_UNROT3L:
|
|
346
|
9498 /* Undo ROT3L: "231" -> "123" */
|
|
330
|
9499 p = fword + sp->ts_fidx;
|
|
|
9500 #ifdef FEAT_MBYTE
|
|
|
9501 if (has_mbyte)
|
|
|
9502 {
|
|
|
9503 n = MB_BYTE2LEN(*p);
|
|
|
9504 n += MB_BYTE2LEN(p[n]);
|
|
|
9505 c = mb_ptr2char(p + n);
|
|
|
9506 tl = MB_BYTE2LEN(p[n]);
|
|
|
9507 mch_memmove(p + tl, p, n);
|
|
|
9508 mb_char2bytes(c, p);
|
|
|
9509 }
|
|
|
9510 else
|
|
|
9511 #endif
|
|
|
9512 {
|
|
|
9513 c = p[2];
|
|
|
9514 p[2] = p[1];
|
|
|
9515 p[1] = *p;
|
|
|
9516 *p = c;
|
|
|
9517 }
|
|
346
|
9518
|
|
323
|
9519 /* Rotate three bytes right: "123" -> "312". We change
|
|
330
|
9520 * "fword" here, it's changed back afterwards. */
|
|
323
|
9521 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
9522 {
|
|
330
|
9523 sp->ts_state = STATE_UNROT3R;
|
|
323
|
9524 ++depth;
|
|
330
|
9525 p = fword + sp->ts_fidx;
|
|
|
9526 #ifdef FEAT_MBYTE
|
|
|
9527 if (has_mbyte)
|
|
|
9528 {
|
|
474
|
9529 n = mb_cptr2len(p);
|
|
|
9530 n += mb_cptr2len(p + n);
|
|
330
|
9531 c = mb_ptr2char(p + n);
|
|
474
|
9532 tl = mb_cptr2len(p + n);
|
|
330
|
9533 mch_memmove(p + tl, p, n);
|
|
|
9534 mb_char2bytes(c, p);
|
|
|
9535 stack[depth].ts_fidxtry = sp->ts_fidx + n + tl;
|
|
|
9536 }
|
|
|
9537 else
|
|
|
9538 #endif
|
|
|
9539 {
|
|
|
9540 c = p[2];
|
|
|
9541 p[2] = p[1];
|
|
|
9542 p[1] = *p;
|
|
|
9543 *p = c;
|
|
|
9544 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
9545 }
|
|
323
|
9546 }
|
|
|
9547 else
|
|
|
9548 sp->ts_state = STATE_REP_INI;
|
|
|
9549 break;
|
|
|
9550
|
|
330
|
9551 case STATE_UNROT3R:
|
|
346
|
9552 /* Undo ROT3R: "312" -> "123" */
|
|
330
|
9553 p = fword + sp->ts_fidx;
|
|
|
9554 #ifdef FEAT_MBYTE
|
|
|
9555 if (has_mbyte)
|
|
|
9556 {
|
|
|
9557 c = mb_ptr2char(p);
|
|
|
9558 tl = MB_BYTE2LEN(*p);
|
|
|
9559 n = MB_BYTE2LEN(p[tl]);
|
|
|
9560 n += MB_BYTE2LEN(p[tl + n]);
|
|
|
9561 mch_memmove(p, p + tl, n);
|
|
|
9562 mb_char2bytes(c, p + n);
|
|
|
9563 }
|
|
|
9564 else
|
|
|
9565 #endif
|
|
|
9566 {
|
|
|
9567 c = *p;
|
|
|
9568 *p = p[1];
|
|
|
9569 p[1] = p[2];
|
|
|
9570 p[2] = c;
|
|
|
9571 }
|
|
323
|
9572 /*FALLTHROUGH*/
|
|
|
9573
|
|
|
9574 case STATE_REP_INI:
|
|
|
9575 /* Check if matching with REP items from the .aff file would
|
|
|
9576 * work. Quickly skip if there are no REP items or the score
|
|
|
9577 * is going to be too high anyway. */
|
|
482
|
9578 gap = &slang->sl_rep;
|
|
323
|
9579 if (gap->ga_len == 0
|
|
|
9580 || sp->ts_score + SCORE_REP >= su->su_maxscore)
|
|
|
9581 {
|
|
|
9582 sp->ts_state = STATE_FINAL;
|
|
|
9583 break;
|
|
|
9584 }
|
|
|
9585
|
|
|
9586 /* Use the first byte to quickly find the first entry that
|
|
330
|
9587 * may match. If the index is -1 there is none. */
|
|
482
|
9588 sp->ts_curi = slang->sl_rep_first[fword[sp->ts_fidx]];
|
|
323
|
9589 if (sp->ts_curi < 0)
|
|
|
9590 {
|
|
|
9591 sp->ts_state = STATE_FINAL;
|
|
|
9592 break;
|
|
|
9593 }
|
|
|
9594
|
|
|
9595 sp->ts_state = STATE_REP;
|
|
|
9596 /*FALLTHROUGH*/
|
|
|
9597
|
|
|
9598 case STATE_REP:
|
|
|
9599 /* Try matching with REP items from the .aff file. For each
|
|
330
|
9600 * match replace the characters and check if the resulting
|
|
|
9601 * word is valid. */
|
|
323
|
9602 p = fword + sp->ts_fidx;
|
|
|
9603
|
|
482
|
9604 gap = &slang->sl_rep;
|
|
323
|
9605 while (sp->ts_curi < gap->ga_len)
|
|
|
9606 {
|
|
|
9607 ftp = (fromto_T *)gap->ga_data + sp->ts_curi++;
|
|
|
9608 if (*ftp->ft_from != *p)
|
|
|
9609 {
|
|
|
9610 /* past possible matching entries */
|
|
|
9611 sp->ts_curi = gap->ga_len;
|
|
|
9612 break;
|
|
|
9613 }
|
|
|
9614 if (STRNCMP(ftp->ft_from, p, STRLEN(ftp->ft_from)) == 0
|
|
|
9615 && try_deeper(su, stack, depth, SCORE_REP))
|
|
|
9616 {
|
|
|
9617 /* Need to undo this afterwards. */
|
|
|
9618 sp->ts_state = STATE_REP_UNDO;
|
|
|
9619
|
|
|
9620 /* Change the "from" to the "to" string. */
|
|
|
9621 ++depth;
|
|
|
9622 fl = STRLEN(ftp->ft_from);
|
|
|
9623 tl = STRLEN(ftp->ft_to);
|
|
|
9624 if (fl != tl)
|
|
346
|
9625 {
|
|
323
|
9626 mch_memmove(p + tl, p + fl, STRLEN(p + fl) + 1);
|
|
346
|
9627 repextra += tl - fl;
|
|
|
9628 }
|
|
323
|
9629 mch_memmove(p, ftp->ft_to, tl);
|
|
|
9630 stack[depth].ts_fidxtry = sp->ts_fidx + tl;
|
|
330
|
9631 #ifdef FEAT_MBYTE
|
|
|
9632 stack[depth].ts_tcharlen = 0;
|
|
|
9633 #endif
|
|
323
|
9634 break;
|
|
|
9635 }
|
|
|
9636 }
|
|
|
9637
|
|
|
9638 if (sp->ts_curi >= gap->ga_len)
|
|
|
9639 /* No (more) matches. */
|
|
|
9640 sp->ts_state = STATE_FINAL;
|
|
|
9641
|
|
|
9642 break;
|
|
|
9643
|
|
|
9644 case STATE_REP_UNDO:
|
|
|
9645 /* Undo a REP replacement and continue with the next one. */
|
|
482
|
9646 ftp = (fromto_T *)slang->sl_rep.ga_data + sp->ts_curi - 1;
|
|
323
|
9647 fl = STRLEN(ftp->ft_from);
|
|
|
9648 tl = STRLEN(ftp->ft_to);
|
|
|
9649 p = fword + sp->ts_fidx;
|
|
|
9650 if (fl != tl)
|
|
346
|
9651 {
|
|
323
|
9652 mch_memmove(p + fl, p + tl, STRLEN(p + tl) + 1);
|
|
346
|
9653 repextra -= tl - fl;
|
|
|
9654 }
|
|
323
|
9655 mch_memmove(p, ftp->ft_from, fl);
|
|
|
9656 sp->ts_state = STATE_REP;
|
|
|
9657 break;
|
|
|
9658
|
|
|
9659 default:
|
|
|
9660 /* Did all possible states at this level, go up one level. */
|
|
|
9661 --depth;
|
|
344
|
9662
|
|
482
|
9663 if (depth >= 0 && stack[depth].ts_prefixdepth == PFD_PREFIXTREE)
|
|
375
|
9664 {
|
|
|
9665 /* Continue in or go back to the prefix tree. */
|
|
|
9666 byts = pbyts;
|
|
|
9667 idxs = pidxs;
|
|
|
9668 }
|
|
|
9669
|
|
344
|
9670 /* Don't check for CTRL-C too often, it takes time. */
|
|
|
9671 line_breakcheck();
|
|
323
|
9672 }
|
|
|
9673 }
|
|
|
9674 }
|
|
|
9675 }
|
|
|
9676
|
|
|
9677 /*
|
|
|
9678 * Try going one level deeper in the tree.
|
|
|
9679 */
|
|
|
9680 static int
|
|
|
9681 try_deeper(su, stack, depth, score_add)
|
|
|
9682 suginfo_T *su;
|
|
|
9683 trystate_T *stack;
|
|
|
9684 int depth;
|
|
|
9685 int score_add;
|
|
|
9686 {
|
|
|
9687 int newscore;
|
|
|
9688
|
|
|
9689 /* Refuse to go deeper if the scrore is getting too big. */
|
|
|
9690 newscore = stack[depth].ts_score + score_add;
|
|
|
9691 if (newscore >= su->su_maxscore)
|
|
|
9692 return FALSE;
|
|
|
9693
|
|
330
|
9694 stack[depth + 1] = stack[depth];
|
|
323
|
9695 stack[depth + 1].ts_state = STATE_START;
|
|
|
9696 stack[depth + 1].ts_score = newscore;
|
|
|
9697 stack[depth + 1].ts_curi = 1; /* start just after length byte */
|
|
485
|
9698 stack[depth + 1].ts_flags = 0;
|
|
323
|
9699 return TRUE;
|
|
|
9700 }
|
|
|
9701
|
|
455
|
9702 #ifdef FEAT_MBYTE
|
|
|
9703 /*
|
|
|
9704 * Case-folding may change the number of bytes: Count nr of chars in
|
|
|
9705 * fword[flen] and return the byte length of that many chars in "word".
|
|
|
9706 */
|
|
|
9707 static int
|
|
|
9708 nofold_len(fword, flen, word)
|
|
|
9709 char_u *fword;
|
|
|
9710 int flen;
|
|
|
9711 char_u *word;
|
|
|
9712 {
|
|
|
9713 char_u *p;
|
|
|
9714 int i = 0;
|
|
|
9715
|
|
|
9716 for (p = fword; p < fword + flen; mb_ptr_adv(p))
|
|
|
9717 ++i;
|
|
|
9718 for (p = word; i > 0; mb_ptr_adv(p))
|
|
|
9719 --i;
|
|
|
9720 return (int)(p - word);
|
|
|
9721 }
|
|
|
9722 #endif
|
|
|
9723
|
|
323
|
9724 /*
|
|
|
9725 * "fword" is a good word with case folded. Find the matching keep-case
|
|
|
9726 * words and put it in "kword".
|
|
|
9727 * Theoretically there could be several keep-case words that result in the
|
|
|
9728 * same case-folded word, but we only find one...
|
|
|
9729 */
|
|
|
9730 static void
|
|
|
9731 find_keepcap_word(slang, fword, kword)
|
|
|
9732 slang_T *slang;
|
|
|
9733 char_u *fword;
|
|
|
9734 char_u *kword;
|
|
|
9735 {
|
|
|
9736 char_u uword[MAXWLEN]; /* "fword" in upper-case */
|
|
|
9737 int depth;
|
|
324
|
9738 idx_T tryidx;
|
|
323
|
9739
|
|
|
9740 /* The following arrays are used at each depth in the tree. */
|
|
324
|
9741 idx_T arridx[MAXWLEN];
|
|
323
|
9742 int round[MAXWLEN];
|
|
|
9743 int fwordidx[MAXWLEN];
|
|
|
9744 int uwordidx[MAXWLEN];
|
|
|
9745 int kwordlen[MAXWLEN];
|
|
|
9746
|
|
|
9747 int flen, ulen;
|
|
|
9748 int l;
|
|
|
9749 int len;
|
|
|
9750 int c;
|
|
324
|
9751 idx_T lo, hi, m;
|
|
323
|
9752 char_u *p;
|
|
|
9753 char_u *byts = slang->sl_kbyts; /* array with bytes of the words */
|
|
324
|
9754 idx_T *idxs = slang->sl_kidxs; /* array with indexes */
|
|
323
|
9755
|
|
|
9756 if (byts == NULL)
|
|
|
9757 {
|
|
|
9758 /* array is empty: "cannot happen" */
|
|
|
9759 *kword = NUL;
|
|
|
9760 return;
|
|
|
9761 }
|
|
|
9762
|
|
|
9763 /* Make an all-cap version of "fword". */
|
|
|
9764 allcap_copy(fword, uword);
|
|
|
9765
|
|
|
9766 /*
|
|
|
9767 * Each character needs to be tried both case-folded and upper-case.
|
|
|
9768 * All this gets very complicated if we keep in mind that changing case
|
|
|
9769 * may change the byte length of a multi-byte character...
|
|
|
9770 */
|
|
|
9771 depth = 0;
|
|
|
9772 arridx[0] = 0;
|
|
|
9773 round[0] = 0;
|
|
|
9774 fwordidx[0] = 0;
|
|
|
9775 uwordidx[0] = 0;
|
|
|
9776 kwordlen[0] = 0;
|
|
|
9777 while (depth >= 0)
|
|
|
9778 {
|
|
|
9779 if (fword[fwordidx[depth]] == NUL)
|
|
|
9780 {
|
|
|
9781 /* We are at the end of "fword". If the tree allows a word to end
|
|
|
9782 * here we have found a match. */
|
|
|
9783 if (byts[arridx[depth] + 1] == 0)
|
|
|
9784 {
|
|
|
9785 kword[kwordlen[depth]] = NUL;
|
|
|
9786 return;
|
|
|
9787 }
|
|
|
9788
|
|
|
9789 /* kword is getting too long, continue one level up */
|
|
|
9790 --depth;
|
|
|
9791 }
|
|
|
9792 else if (++round[depth] > 2)
|
|
|
9793 {
|
|
|
9794 /* tried both fold-case and upper-case character, continue one
|
|
|
9795 * level up */
|
|
|
9796 --depth;
|
|
|
9797 }
|
|
|
9798 else
|
|
|
9799 {
|
|
|
9800 /*
|
|
|
9801 * round[depth] == 1: Try using the folded-case character.
|
|
|
9802 * round[depth] == 2: Try using the upper-case character.
|
|
|
9803 */
|
|
|
9804 #ifdef FEAT_MBYTE
|
|
|
9805 if (has_mbyte)
|
|
|
9806 {
|
|
474
|
9807 flen = mb_cptr2len(fword + fwordidx[depth]);
|
|
|
9808 ulen = mb_cptr2len(uword + uwordidx[depth]);
|
|
323
|
9809 }
|
|
|
9810 else
|
|
|
9811 #endif
|
|
|
9812 ulen = flen = 1;
|
|
|
9813 if (round[depth] == 1)
|
|
|
9814 {
|
|
|
9815 p = fword + fwordidx[depth];
|
|
|
9816 l = flen;
|
|
|
9817 }
|
|
|
9818 else
|
|
|
9819 {
|
|
|
9820 p = uword + uwordidx[depth];
|
|
|
9821 l = ulen;
|
|
|
9822 }
|
|
|
9823
|
|
|
9824 for (tryidx = arridx[depth]; l > 0; --l)
|
|
|
9825 {
|
|
|
9826 /* Perform a binary search in the list of accepted bytes. */
|
|
|
9827 len = byts[tryidx++];
|
|
|
9828 c = *p++;
|
|
|
9829 lo = tryidx;
|
|
|
9830 hi = tryidx + len - 1;
|
|
|
9831 while (lo < hi)
|
|
|
9832 {
|
|
|
9833 m = (lo + hi) / 2;
|
|
|
9834 if (byts[m] > c)
|
|
|
9835 hi = m - 1;
|
|
|
9836 else if (byts[m] < c)
|
|
|
9837 lo = m + 1;
|
|
|
9838 else
|
|
|
9839 {
|
|
|
9840 lo = hi = m;
|
|
|
9841 break;
|
|
|
9842 }
|
|
|
9843 }
|
|
|
9844
|
|
|
9845 /* Stop if there is no matching byte. */
|
|
|
9846 if (hi < lo || byts[lo] != c)
|
|
|
9847 break;
|
|
|
9848
|
|
|
9849 /* Continue at the child (if there is one). */
|
|
|
9850 tryidx = idxs[lo];
|
|
|
9851 }
|
|
|
9852
|
|
|
9853 if (l == 0)
|
|
|
9854 {
|
|
|
9855 /*
|
|
|
9856 * Found the matching char. Copy it to "kword" and go a
|
|
|
9857 * level deeper.
|
|
|
9858 */
|
|
|
9859 if (round[depth] == 1)
|
|
|
9860 {
|
|
|
9861 STRNCPY(kword + kwordlen[depth], fword + fwordidx[depth],
|
|
|
9862 flen);
|
|
|
9863 kwordlen[depth + 1] = kwordlen[depth] + flen;
|
|
|
9864 }
|
|
|
9865 else
|
|
|
9866 {
|
|
|
9867 STRNCPY(kword + kwordlen[depth], uword + uwordidx[depth],
|
|
|
9868 ulen);
|
|
|
9869 kwordlen[depth + 1] = kwordlen[depth] + ulen;
|
|
|
9870 }
|
|
|
9871 fwordidx[depth + 1] = fwordidx[depth] + flen;
|
|
|
9872 uwordidx[depth + 1] = uwordidx[depth] + ulen;
|
|
|
9873
|
|
|
9874 ++depth;
|
|
|
9875 arridx[depth] = tryidx;
|
|
|
9876 round[depth] = 0;
|
|
|
9877 }
|
|
|
9878 }
|
|
|
9879 }
|
|
|
9880
|
|
|
9881 /* Didn't find it: "cannot happen". */
|
|
|
9882 *kword = NUL;
|
|
|
9883 }
|
|
|
9884
|
|
|
9885 /*
|
|
344
|
9886 * Compute the sound-a-like score for suggestions in su->su_ga and add them to
|
|
|
9887 * su->su_sga.
|
|
|
9888 */
|
|
|
9889 static void
|
|
|
9890 score_comp_sal(su)
|
|
|
9891 suginfo_T *su;
|
|
|
9892 {
|
|
|
9893 langp_T *lp;
|
|
|
9894 char_u badsound[MAXWLEN];
|
|
|
9895 int i;
|
|
|
9896 suggest_T *stp;
|
|
|
9897 suggest_T *sstp;
|
|
|
9898 int score;
|
|
|
9899
|
|
|
9900 if (ga_grow(&su->su_sga, su->su_ga.ga_len) == FAIL)
|
|
|
9901 return;
|
|
|
9902
|
|
|
9903 /* Use the sound-folding of the first language that supports it. */
|
|
|
9904 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
9905 lp->lp_slang != NULL; ++lp)
|
|
|
9906 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
9907 {
|
|
|
9908 /* soundfold the bad word */
|
|
375
|
9909 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound);
|
|
344
|
9910
|
|
|
9911 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
9912 {
|
|
|
9913 stp = &SUG(su->su_ga, i);
|
|
|
9914
|
|
351
|
9915 /* Case-fold the suggested word, sound-fold it and compute the
|
|
|
9916 * sound-a-like score. */
|
|
|
9917 score = stp_sal_score(stp, su, lp->lp_slang, badsound);
|
|
344
|
9918 if (score < SCORE_MAXMAX)
|
|
|
9919 {
|
|
|
9920 /* Add the suggestion. */
|
|
|
9921 sstp = &SUG(su->su_sga, su->su_sga.ga_len);
|
|
|
9922 sstp->st_word = vim_strsave(stp->st_word);
|
|
|
9923 if (sstp->st_word != NULL)
|
|
|
9924 {
|
|
|
9925 sstp->st_score = score;
|
|
|
9926 sstp->st_altscore = 0;
|
|
|
9927 sstp->st_orglen = stp->st_orglen;
|
|
|
9928 ++su->su_sga.ga_len;
|
|
|
9929 }
|
|
|
9930 }
|
|
|
9931 }
|
|
|
9932 break;
|
|
|
9933 }
|
|
|
9934 }
|
|
|
9935
|
|
|
9936 /*
|
|
|
9937 * Combine the list of suggestions in su->su_ga and su->su_sga.
|
|
|
9938 * They are intwined.
|
|
|
9939 */
|
|
|
9940 static void
|
|
|
9941 score_combine(su)
|
|
|
9942 suginfo_T *su;
|
|
|
9943 {
|
|
|
9944 int i;
|
|
|
9945 int j;
|
|
|
9946 garray_T ga;
|
|
|
9947 garray_T *gap;
|
|
|
9948 langp_T *lp;
|
|
|
9949 suggest_T *stp;
|
|
|
9950 char_u *p;
|
|
|
9951 char_u badsound[MAXWLEN];
|
|
|
9952 int round;
|
|
|
9953
|
|
|
9954 /* Add the alternate score to su_ga. */
|
|
|
9955 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
9956 lp->lp_slang != NULL; ++lp)
|
|
|
9957 {
|
|
|
9958 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
9959 {
|
|
|
9960 /* soundfold the bad word */
|
|
375
|
9961 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound);
|
|
344
|
9962
|
|
|
9963 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
9964 {
|
|
|
9965 stp = &SUG(su->su_ga, i);
|
|
351
|
9966 stp->st_altscore = stp_sal_score(stp, su, lp->lp_slang,
|
|
|
9967 badsound);
|
|
344
|
9968 if (stp->st_altscore == SCORE_MAXMAX)
|
|
|
9969 stp->st_score = (stp->st_score * 3 + SCORE_BIG) / 4;
|
|
|
9970 else
|
|
|
9971 stp->st_score = (stp->st_score * 3
|
|
|
9972 + stp->st_altscore) / 4;
|
|
|
9973 stp->st_salscore = FALSE;
|
|
|
9974 }
|
|
|
9975 break;
|
|
|
9976 }
|
|
|
9977 }
|
|
|
9978
|
|
|
9979 /* Add the alternate score to su_sga. */
|
|
|
9980 for (i = 0; i < su->su_sga.ga_len; ++i)
|
|
|
9981 {
|
|
|
9982 stp = &SUG(su->su_sga, i);
|
|
|
9983 stp->st_altscore = spell_edit_score(su->su_badword, stp->st_word);
|
|
|
9984 if (stp->st_score == SCORE_MAXMAX)
|
|
|
9985 stp->st_score = (SCORE_BIG * 7 + stp->st_altscore) / 8;
|
|
|
9986 else
|
|
|
9987 stp->st_score = (stp->st_score * 7 + stp->st_altscore) / 8;
|
|
|
9988 stp->st_salscore = TRUE;
|
|
|
9989 }
|
|
|
9990
|
|
|
9991 /* Sort the suggestions and truncate at "maxcount" for both lists. */
|
|
|
9992 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
9993 (void)cleanup_suggestions(&su->su_sga, su->su_maxscore, su->su_maxcount);
|
|
|
9994
|
|
|
9995 ga_init2(&ga, (int)sizeof(suginfo_T), 1);
|
|
|
9996 if (ga_grow(&ga, su->su_ga.ga_len + su->su_sga.ga_len) == FAIL)
|
|
|
9997 return;
|
|
|
9998
|
|
|
9999 stp = &SUG(ga, 0);
|
|
|
10000 for (i = 0; i < su->su_ga.ga_len || i < su->su_sga.ga_len; ++i)
|
|
|
10001 {
|
|
|
10002 /* round 1: get a suggestion from su_ga
|
|
|
10003 * round 2: get a suggestion from su_sga */
|
|
|
10004 for (round = 1; round <= 2; ++round)
|
|
|
10005 {
|
|
|
10006 gap = round == 1 ? &su->su_ga : &su->su_sga;
|
|
|
10007 if (i < gap->ga_len)
|
|
|
10008 {
|
|
|
10009 /* Don't add a word if it's already there. */
|
|
|
10010 p = SUG(*gap, i).st_word;
|
|
|
10011 for (j = 0; j < ga.ga_len; ++j)
|
|
|
10012 if (STRCMP(stp[j].st_word, p) == 0)
|
|
|
10013 break;
|
|
|
10014 if (j == ga.ga_len)
|
|
|
10015 stp[ga.ga_len++] = SUG(*gap, i);
|
|
|
10016 else
|
|
|
10017 vim_free(p);
|
|
|
10018 }
|
|
|
10019 }
|
|
|
10020 }
|
|
|
10021
|
|
|
10022 ga_clear(&su->su_ga);
|
|
|
10023 ga_clear(&su->su_sga);
|
|
|
10024
|
|
|
10025 /* Truncate the list to the number of suggestions that will be displayed. */
|
|
|
10026 if (ga.ga_len > su->su_maxcount)
|
|
|
10027 {
|
|
|
10028 for (i = su->su_maxcount; i < ga.ga_len; ++i)
|
|
|
10029 vim_free(stp[i].st_word);
|
|
|
10030 ga.ga_len = su->su_maxcount;
|
|
|
10031 }
|
|
|
10032
|
|
|
10033 su->su_ga = ga;
|
|
|
10034 }
|
|
|
10035
|
|
|
10036 /*
|
|
351
|
10037 * For the goodword in "stp" compute the soundalike score compared to the
|
|
|
10038 * badword.
|
|
|
10039 */
|
|
|
10040 static int
|
|
|
10041 stp_sal_score(stp, su, slang, badsound)
|
|
|
10042 suggest_T *stp;
|
|
|
10043 suginfo_T *su;
|
|
|
10044 slang_T *slang;
|
|
|
10045 char_u *badsound; /* sound-folded badword */
|
|
|
10046 {
|
|
|
10047 char_u *p;
|
|
|
10048 char_u badsound2[MAXWLEN];
|
|
|
10049 char_u fword[MAXWLEN];
|
|
|
10050 char_u goodsound[MAXWLEN];
|
|
|
10051
|
|
|
10052 if (stp->st_orglen <= su->su_badlen)
|
|
|
10053 p = badsound;
|
|
|
10054 else
|
|
|
10055 {
|
|
|
10056 /* soundfold the bad word with more characters following */
|
|
|
10057 (void)spell_casefold(su->su_badptr, stp->st_orglen, fword, MAXWLEN);
|
|
|
10058
|
|
|
10059 /* When joining two words the sound often changes a lot. E.g., "t he"
|
|
|
10060 * sounds like "t h" while "the" sounds like "@". Avoid that by
|
|
|
10061 * removing the space. Don't do it when the good word also contains a
|
|
|
10062 * space. */
|
|
|
10063 if (vim_iswhite(su->su_badptr[su->su_badlen])
|
|
|
10064 && *skiptowhite(stp->st_word) == NUL)
|
|
|
10065 for (p = fword; *(p = skiptowhite(p)) != NUL; )
|
|
|
10066 mch_memmove(p, p + 1, STRLEN(p));
|
|
|
10067
|
|
375
|
10068 spell_soundfold(slang, fword, TRUE, badsound2);
|
|
351
|
10069 p = badsound2;
|
|
|
10070 }
|
|
|
10071
|
|
375
|
10072 /* Sound-fold the word and compute the score for the difference. */
|
|
|
10073 spell_soundfold(slang, stp->st_word, FALSE, goodsound);
|
|
351
|
10074
|
|
|
10075 return soundalike_score(goodsound, p);
|
|
|
10076 }
|
|
|
10077
|
|
|
10078 /*
|
|
323
|
10079 * Find suggestions by comparing the word in a sound-a-like form.
|
|
|
10080 */
|
|
|
10081 static void
|
|
346
|
10082 suggest_try_soundalike(su)
|
|
323
|
10083 suginfo_T *su;
|
|
|
10084 {
|
|
|
10085 char_u salword[MAXWLEN];
|
|
|
10086 char_u tword[MAXWLEN];
|
|
|
10087 char_u tsalword[MAXWLEN];
|
|
324
|
10088 idx_T arridx[MAXWLEN];
|
|
323
|
10089 int curi[MAXWLEN];
|
|
|
10090 langp_T *lp;
|
|
|
10091 char_u *byts;
|
|
324
|
10092 idx_T *idxs;
|
|
323
|
10093 int depth;
|
|
|
10094 int c;
|
|
324
|
10095 idx_T n;
|
|
323
|
10096 int round;
|
|
|
10097 int flags;
|
|
344
|
10098 int sound_score;
|
|
483
|
10099 int local_score;
|
|
344
|
10100
|
|
|
10101 /* Do this for all languages that support sound folding. */
|
|
323
|
10102 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
10103 lp->lp_slang != NULL; ++lp)
|
|
|
10104 {
|
|
|
10105 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
10106 {
|
|
|
10107 /* soundfold the bad word */
|
|
375
|
10108 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, salword);
|
|
323
|
10109
|
|
|
10110 /*
|
|
|
10111 * Go through the whole tree, soundfold each word and compare.
|
|
|
10112 * round 1: use the case-folded tree.
|
|
|
10113 * round 2: use the keep-case tree.
|
|
|
10114 */
|
|
|
10115 for (round = 1; round <= 2; ++round)
|
|
|
10116 {
|
|
|
10117 if (round == 1)
|
|
|
10118 {
|
|
|
10119 byts = lp->lp_slang->sl_fbyts;
|
|
|
10120 idxs = lp->lp_slang->sl_fidxs;
|
|
|
10121 }
|
|
|
10122 else
|
|
|
10123 {
|
|
|
10124 byts = lp->lp_slang->sl_kbyts;
|
|
|
10125 idxs = lp->lp_slang->sl_kidxs;
|
|
389
|
10126 if (byts == NULL) /* no keep-case words */
|
|
|
10127 continue;
|
|
323
|
10128 }
|
|
|
10129
|
|
|
10130 depth = 0;
|
|
|
10131 arridx[0] = 0;
|
|
|
10132 curi[0] = 1;
|
|
|
10133 while (depth >= 0 && !got_int)
|
|
|
10134 {
|
|
|
10135 if (curi[depth] > byts[arridx[depth]])
|
|
351
|
10136 {
|
|
323
|
10137 /* Done all bytes at this node, go up one level. */
|
|
|
10138 --depth;
|
|
351
|
10139 line_breakcheck();
|
|
|
10140 }
|
|
323
|
10141 else
|
|
|
10142 {
|
|
|
10143 /* Do one more byte at this node. */
|
|
|
10144 n = arridx[depth] + curi[depth];
|
|
|
10145 ++curi[depth];
|
|
|
10146 c = byts[n];
|
|
|
10147 if (c == 0)
|
|
|
10148 {
|
|
|
10149 /* End of word, deal with the word. */
|
|
324
|
10150 flags = (int)idxs[n];
|
|
323
|
10151 if (round == 2 || (flags & WF_KEEPCAP) == 0)
|
|
|
10152 {
|
|
|
10153 tword[depth] = NUL;
|
|
375
|
10154 /* Sound-fold. Only in keep-case tree need to
|
|
|
10155 * case-fold the word. */
|
|
|
10156 spell_soundfold(lp->lp_slang, tword,
|
|
|
10157 round == 1, tsalword);
|
|
323
|
10158
|
|
344
|
10159 /* Compute the edit distance between the
|
|
|
10160 * sound-a-like words. */
|
|
|
10161 sound_score = soundalike_score(salword,
|
|
|
10162 tsalword);
|
|
483
|
10163
|
|
|
10164 /* Add a penalty for words in another region. */
|
|
|
10165 if ((flags & WF_REGION) && (((unsigned)flags
|
|
|
10166 >> 16) & lp->lp_region) == 0)
|
|
|
10167 local_score = SCORE_REGION;
|
|
|
10168 else
|
|
|
10169 local_score = 0;
|
|
|
10170 sound_score += local_score;
|
|
|
10171
|
|
324
|
10172 if (sound_score < SCORE_MAXMAX)
|
|
323
|
10173 {
|
|
324
|
10174 char_u cword[MAXWLEN];
|
|
|
10175 char_u *p;
|
|
344
|
10176 int score;
|
|
324
|
10177
|
|
388
|
10178 flags |= su->su_badflags;
|
|
351
|
10179 if (round == 1 && (flags & WF_CAPMASK) != 0)
|
|
323
|
10180 {
|
|
324
|
10181 /* Need to fix case according to
|
|
|
10182 * "flags". */
|
|
323
|
10183 make_case_word(tword, cword, flags);
|
|
324
|
10184 p = cword;
|
|
323
|
10185 }
|
|
|
10186 else
|
|
324
|
10187 p = tword;
|
|
|
10188
|
|
344
|
10189 if (sps_flags & SPS_DOUBLE)
|
|
|
10190 add_suggestion(su, &su->su_sga, p,
|
|
346
|
10191 su->su_badlen,
|
|
351
|
10192 sound_score, 0, FALSE);
|
|
344
|
10193 else
|
|
|
10194 {
|
|
|
10195 /* Compute the score. */
|
|
|
10196 score = spell_edit_score(
|
|
483
|
10197 su->su_badword, p)
|
|
|
10198 + local_score;
|
|
344
|
10199 if (sps_flags & SPS_BEST)
|
|
|
10200 /* give a bonus for the good word
|
|
|
10201 * sounding the same as the bad
|
|
|
10202 * word */
|
|
|
10203 add_suggestion(su, &su->su_ga, p,
|
|
346
|
10204 su->su_badlen,
|
|
344
|
10205 RESCORE(score, sound_score),
|
|
351
|
10206 sound_score, TRUE);
|
|
344
|
10207 else
|
|
|
10208 add_suggestion(su, &su->su_ga, p,
|
|
346
|
10209 su->su_badlen,
|
|
351
|
10210 score + sound_score, 0, FALSE);
|
|
344
|
10211 }
|
|
323
|
10212 }
|
|
|
10213 }
|
|
|
10214
|
|
|
10215 /* Skip over other NUL bytes. */
|
|
|
10216 while (byts[n + 1] == 0)
|
|
|
10217 {
|
|
|
10218 ++n;
|
|
|
10219 ++curi[depth];
|
|
|
10220 }
|
|
|
10221 }
|
|
|
10222 else
|
|
|
10223 {
|
|
|
10224 /* Normal char, go one level deeper. */
|
|
|
10225 tword[depth++] = c;
|
|
|
10226 arridx[depth] = idxs[n];
|
|
|
10227 curi[depth] = 1;
|
|
|
10228 }
|
|
|
10229 }
|
|
|
10230 }
|
|
|
10231 }
|
|
|
10232 }
|
|
|
10233 }
|
|
|
10234 }
|
|
|
10235
|
|
|
10236 /*
|
|
324
|
10237 * Copy "fword" to "cword", fixing case according to "flags".
|
|
323
|
10238 */
|
|
|
10239 static void
|
|
|
10240 make_case_word(fword, cword, flags)
|
|
|
10241 char_u *fword;
|
|
|
10242 char_u *cword;
|
|
|
10243 int flags;
|
|
|
10244 {
|
|
|
10245 if (flags & WF_ALLCAP)
|
|
|
10246 /* Make it all upper-case */
|
|
|
10247 allcap_copy(fword, cword);
|
|
|
10248 else if (flags & WF_ONECAP)
|
|
|
10249 /* Make the first letter upper-case */
|
|
324
|
10250 onecap_copy(fword, cword, TRUE);
|
|
323
|
10251 else
|
|
|
10252 /* Use goodword as-is. */
|
|
|
10253 STRCPY(cword, fword);
|
|
|
10254 }
|
|
|
10255
|
|
330
|
10256 /*
|
|
|
10257 * Use map string "map" for languages "lp".
|
|
|
10258 */
|
|
|
10259 static void
|
|
|
10260 set_map_str(lp, map)
|
|
|
10261 slang_T *lp;
|
|
|
10262 char_u *map;
|
|
|
10263 {
|
|
|
10264 char_u *p;
|
|
|
10265 int headc = 0;
|
|
|
10266 int c;
|
|
|
10267 int i;
|
|
|
10268
|
|
|
10269 if (*map == NUL)
|
|
|
10270 {
|
|
|
10271 lp->sl_has_map = FALSE;
|
|
|
10272 return;
|
|
|
10273 }
|
|
|
10274 lp->sl_has_map = TRUE;
|
|
|
10275
|
|
|
10276 /* Init the array and hash table empty. */
|
|
|
10277 for (i = 0; i < 256; ++i)
|
|
|
10278 lp->sl_map_array[i] = 0;
|
|
|
10279 #ifdef FEAT_MBYTE
|
|
|
10280 hash_init(&lp->sl_map_hash);
|
|
|
10281 #endif
|
|
|
10282
|
|
|
10283 /*
|
|
|
10284 * The similar characters are stored separated with slashes:
|
|
|
10285 * "aaa/bbb/ccc/". Fill sl_map_array[c] with the character before c and
|
|
|
10286 * before the same slash. For characters above 255 sl_map_hash is used.
|
|
|
10287 */
|
|
|
10288 for (p = map; *p != NUL; )
|
|
|
10289 {
|
|
|
10290 #ifdef FEAT_MBYTE
|
|
474
|
10291 c = mb_cptr2char_adv(&p);
|
|
330
|
10292 #else
|
|
|
10293 c = *p++;
|
|
|
10294 #endif
|
|
|
10295 if (c == '/')
|
|
|
10296 headc = 0;
|
|
|
10297 else
|
|
|
10298 {
|
|
|
10299 if (headc == 0)
|
|
|
10300 headc = c;
|
|
|
10301
|
|
|
10302 #ifdef FEAT_MBYTE
|
|
|
10303 /* Characters above 255 don't fit in sl_map_array[], put them in
|
|
|
10304 * the hash table. Each entry is the char, a NUL the headchar and
|
|
|
10305 * a NUL. */
|
|
|
10306 if (c >= 256)
|
|
|
10307 {
|
|
|
10308 int cl = mb_char2len(c);
|
|
|
10309 int headcl = mb_char2len(headc);
|
|
|
10310 char_u *b;
|
|
|
10311 hash_T hash;
|
|
|
10312 hashitem_T *hi;
|
|
|
10313
|
|
|
10314 b = alloc((unsigned)(cl + headcl + 2));
|
|
|
10315 if (b == NULL)
|
|
|
10316 return;
|
|
|
10317 mb_char2bytes(c, b);
|
|
|
10318 b[cl] = NUL;
|
|
|
10319 mb_char2bytes(headc, b + cl + 1);
|
|
|
10320 b[cl + 1 + headcl] = NUL;
|
|
|
10321 hash = hash_hash(b);
|
|
|
10322 hi = hash_lookup(&lp->sl_map_hash, b, hash);
|
|
|
10323 if (HASHITEM_EMPTY(hi))
|
|
|
10324 hash_add_item(&lp->sl_map_hash, hi, b, hash);
|
|
|
10325 else
|
|
|
10326 {
|
|
|
10327 /* This should have been checked when generating the .spl
|
|
|
10328 * file. */
|
|
|
10329 EMSG(_("E999: duplicate char in MAP entry"));
|
|
|
10330 vim_free(b);
|
|
|
10331 }
|
|
|
10332 }
|
|
|
10333 else
|
|
|
10334 #endif
|
|
|
10335 lp->sl_map_array[c] = headc;
|
|
|
10336 }
|
|
|
10337 }
|
|
|
10338 }
|
|
|
10339
|
|
323
|
10340 /*
|
|
|
10341 * Return TRUE if "c1" and "c2" are similar characters according to the MAP
|
|
|
10342 * lines in the .aff file.
|
|
|
10343 */
|
|
|
10344 static int
|
|
|
10345 similar_chars(slang, c1, c2)
|
|
|
10346 slang_T *slang;
|
|
|
10347 int c1;
|
|
|
10348 int c2;
|
|
|
10349 {
|
|
330
|
10350 int m1, m2;
|
|
|
10351 #ifdef FEAT_MBYTE
|
|
|
10352 char_u buf[MB_MAXBYTES];
|
|
|
10353 hashitem_T *hi;
|
|
|
10354
|
|
|
10355 if (c1 >= 256)
|
|
|
10356 {
|
|
|
10357 buf[mb_char2bytes(c1, buf)] = 0;
|
|
|
10358 hi = hash_find(&slang->sl_map_hash, buf);
|
|
|
10359 if (HASHITEM_EMPTY(hi))
|
|
|
10360 m1 = 0;
|
|
|
10361 else
|
|
|
10362 m1 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1);
|
|
|
10363 }
|
|
|
10364 else
|
|
|
10365 #endif
|
|
|
10366 m1 = slang->sl_map_array[c1];
|
|
|
10367 if (m1 == 0)
|
|
323
|
10368 return FALSE;
|
|
330
|
10369
|
|
|
10370
|
|
|
10371 #ifdef FEAT_MBYTE
|
|
|
10372 if (c2 >= 256)
|
|
|
10373 {
|
|
|
10374 buf[mb_char2bytes(c2, buf)] = 0;
|
|
|
10375 hi = hash_find(&slang->sl_map_hash, buf);
|
|
|
10376 if (HASHITEM_EMPTY(hi))
|
|
|
10377 m2 = 0;
|
|
|
10378 else
|
|
|
10379 m2 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1);
|
|
|
10380 }
|
|
|
10381 else
|
|
|
10382 #endif
|
|
|
10383 m2 = slang->sl_map_array[c2];
|
|
|
10384
|
|
|
10385 return m1 == m2;
|
|
323
|
10386 }
|
|
|
10387
|
|
|
10388 /*
|
|
|
10389 * Add a suggestion to the list of suggestions.
|
|
|
10390 * Do not add a duplicate suggestion or suggestions with a bad score.
|
|
|
10391 * When "use_score" is not zero it's used, otherwise the score is computed
|
|
|
10392 * with spell_edit_score().
|
|
|
10393 */
|
|
|
10394 static void
|
|
351
|
10395 add_suggestion(su, gap, goodword, badlen, score, altscore, had_bonus)
|
|
323
|
10396 suginfo_T *su;
|
|
344
|
10397 garray_T *gap;
|
|
323
|
10398 char_u *goodword;
|
|
346
|
10399 int badlen; /* length of bad word used */
|
|
324
|
10400 int score;
|
|
351
|
10401 int altscore;
|
|
344
|
10402 int had_bonus; /* value for st_had_bonus */
|
|
323
|
10403 {
|
|
|
10404 suggest_T *stp;
|
|
|
10405 int i;
|
|
346
|
10406 char_u *p = NULL;
|
|
|
10407 int c = 0;
|
|
323
|
10408
|
|
|
10409 /* Check that the word wasn't banned. */
|
|
|
10410 if (was_banned(su, goodword))
|
|
|
10411 return;
|
|
|
10412
|
|
346
|
10413 /* If past "su_badlen" and the rest is identical stop at "su_badlen".
|
|
|
10414 * Remove the common part from "goodword". */
|
|
|
10415 i = badlen - su->su_badlen;
|
|
|
10416 if (i > 0)
|
|
|
10417 {
|
|
|
10418 /* This assumes there was no case folding or it didn't change the
|
|
|
10419 * length... */
|
|
|
10420 p = goodword + STRLEN(goodword) - i;
|
|
|
10421 if (p > goodword && STRNICMP(su->su_badptr + su->su_badlen, p, i) == 0)
|
|
|
10422 {
|
|
|
10423 badlen = su->su_badlen;
|
|
|
10424 c = *p;
|
|
|
10425 *p = NUL;
|
|
|
10426 }
|
|
|
10427 else
|
|
|
10428 p = NULL;
|
|
|
10429 }
|
|
372
|
10430 else if (i < 0)
|
|
|
10431 {
|
|
|
10432 /* When replacing part of the word check that we actually change
|
|
|
10433 * something. For "the the" a suggestion can be replacing the first
|
|
|
10434 * "the" with itself, since "the" wasn't banned. */
|
|
376
|
10435 if (badlen == (int)STRLEN(goodword)
|
|
372
|
10436 && STRNCMP(su->su_badword, goodword, badlen) == 0)
|
|
|
10437 return;
|
|
|
10438 }
|
|
|
10439
|
|
346
|
10440
|
|
323
|
10441 if (score <= su->su_maxscore)
|
|
|
10442 {
|
|
366
|
10443 /* Check if the word is already there. Also check the length that is
|
|
|
10444 * being replaced "thes," -> "these" is a different suggestion from
|
|
|
10445 * "thes" -> "these". */
|
|
344
|
10446 stp = &SUG(*gap, 0);
|
|
|
10447 for (i = gap->ga_len - 1; i >= 0; --i)
|
|
366
|
10448 if (STRCMP(stp[i].st_word, goodword) == 0
|
|
|
10449 && stp[i].st_orglen == badlen)
|
|
323
|
10450 {
|
|
|
10451 /* Found it. Remember the lowest score. */
|
|
|
10452 if (stp[i].st_score > score)
|
|
324
|
10453 {
|
|
323
|
10454 stp[i].st_score = score;
|
|
376
|
10455 stp[i].st_altscore = altscore;
|
|
324
|
10456 stp[i].st_had_bonus = had_bonus;
|
|
|
10457 }
|
|
323
|
10458 break;
|
|
|
10459 }
|
|
|
10460
|
|
344
|
10461 if (i < 0 && ga_grow(gap, 1) == OK)
|
|
323
|
10462 {
|
|
|
10463 /* Add a suggestion. */
|
|
344
|
10464 stp = &SUG(*gap, gap->ga_len);
|
|
323
|
10465 stp->st_word = vim_strsave(goodword);
|
|
|
10466 if (stp->st_word != NULL)
|
|
|
10467 {
|
|
|
10468 stp->st_score = score;
|
|
351
|
10469 stp->st_altscore = altscore;
|
|
324
|
10470 stp->st_had_bonus = had_bonus;
|
|
346
|
10471 stp->st_orglen = badlen;
|
|
344
|
10472 ++gap->ga_len;
|
|
323
|
10473
|
|
|
10474 /* If we have too many suggestions now, sort the list and keep
|
|
|
10475 * the best suggestions. */
|
|
344
|
10476 if (gap->ga_len > SUG_MAX_COUNT(su))
|
|
|
10477 su->su_maxscore = cleanup_suggestions(gap, su->su_maxscore,
|
|
|
10478 SUG_CLEAN_COUNT(su));
|
|
323
|
10479 }
|
|
|
10480 }
|
|
|
10481 }
|
|
346
|
10482
|
|
|
10483 if (p != NULL)
|
|
|
10484 *p = c; /* restore "goodword" */
|
|
323
|
10485 }
|
|
|
10486
|
|
|
10487 /*
|
|
|
10488 * Add a word to be banned.
|
|
|
10489 */
|
|
|
10490 static void
|
|
|
10491 add_banned(su, word)
|
|
|
10492 suginfo_T *su;
|
|
|
10493 char_u *word;
|
|
|
10494 {
|
|
|
10495 char_u *s = vim_strsave(word);
|
|
|
10496 hash_T hash;
|
|
|
10497 hashitem_T *hi;
|
|
|
10498
|
|
|
10499 if (s != NULL)
|
|
|
10500 {
|
|
|
10501 hash = hash_hash(s);
|
|
|
10502 hi = hash_lookup(&su->su_banned, s, hash);
|
|
|
10503 if (HASHITEM_EMPTY(hi))
|
|
|
10504 hash_add_item(&su->su_banned, hi, s, hash);
|
|
355
|
10505 else
|
|
|
10506 vim_free(s);
|
|
323
|
10507 }
|
|
|
10508 }
|
|
|
10509
|
|
|
10510 /*
|
|
|
10511 * Return TRUE if a word appears in the list of banned words.
|
|
|
10512 */
|
|
|
10513 static int
|
|
|
10514 was_banned(su, word)
|
|
|
10515 suginfo_T *su;
|
|
|
10516 char_u *word;
|
|
|
10517 {
|
|
324
|
10518 hashitem_T *hi = hash_find(&su->su_banned, word);
|
|
|
10519
|
|
|
10520 return !HASHITEM_EMPTY(hi);
|
|
323
|
10521 }
|
|
|
10522
|
|
|
10523 /*
|
|
|
10524 * Free the banned words in "su".
|
|
|
10525 */
|
|
|
10526 static void
|
|
|
10527 free_banned(su)
|
|
|
10528 suginfo_T *su;
|
|
|
10529 {
|
|
|
10530 int todo;
|
|
|
10531 hashitem_T *hi;
|
|
|
10532
|
|
|
10533 todo = su->su_banned.ht_used;
|
|
|
10534 for (hi = su->su_banned.ht_array; todo > 0; ++hi)
|
|
|
10535 {
|
|
|
10536 if (!HASHITEM_EMPTY(hi))
|
|
|
10537 {
|
|
|
10538 vim_free(hi->hi_key);
|
|
|
10539 --todo;
|
|
|
10540 }
|
|
|
10541 }
|
|
|
10542 hash_clear(&su->su_banned);
|
|
|
10543 }
|
|
|
10544
|
|
324
|
10545 /*
|
|
|
10546 * Recompute the score if sound-folding is possible. This is slow,
|
|
|
10547 * thus only done for the final results.
|
|
|
10548 */
|
|
|
10549 static void
|
|
|
10550 rescore_suggestions(su)
|
|
|
10551 suginfo_T *su;
|
|
|
10552 {
|
|
|
10553 langp_T *lp;
|
|
|
10554 suggest_T *stp;
|
|
|
10555 char_u sal_badword[MAXWLEN];
|
|
|
10556 int i;
|
|
|
10557
|
|
|
10558 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
10559 lp->lp_slang != NULL; ++lp)
|
|
|
10560 {
|
|
|
10561 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
10562 {
|
|
|
10563 /* soundfold the bad word */
|
|
375
|
10564 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, sal_badword);
|
|
324
|
10565
|
|
|
10566 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
10567 {
|
|
344
|
10568 stp = &SUG(su->su_ga, i);
|
|
324
|
10569 if (!stp->st_had_bonus)
|
|
|
10570 {
|
|
351
|
10571 stp->st_altscore = stp_sal_score(stp, su,
|
|
|
10572 lp->lp_slang, sal_badword);
|
|
|
10573 if (stp->st_altscore == SCORE_MAXMAX)
|
|
|
10574 stp->st_altscore = SCORE_BIG;
|
|
|
10575 stp->st_score = RESCORE(stp->st_score, stp->st_altscore);
|
|
324
|
10576 }
|
|
|
10577 }
|
|
|
10578 break;
|
|
|
10579 }
|
|
|
10580 }
|
|
|
10581 }
|
|
|
10582
|
|
323
|
10583 static int
|
|
|
10584 #ifdef __BORLANDC__
|
|
|
10585 _RTLENTRYF
|
|
|
10586 #endif
|
|
|
10587 sug_compare __ARGS((const void *s1, const void *s2));
|
|
|
10588
|
|
|
10589 /*
|
|
|
10590 * Function given to qsort() to sort the suggestions on st_score.
|
|
|
10591 */
|
|
|
10592 static int
|
|
|
10593 #ifdef __BORLANDC__
|
|
|
10594 _RTLENTRYF
|
|
|
10595 #endif
|
|
|
10596 sug_compare(s1, s2)
|
|
|
10597 const void *s1;
|
|
|
10598 const void *s2;
|
|
|
10599 {
|
|
|
10600 suggest_T *p1 = (suggest_T *)s1;
|
|
|
10601 suggest_T *p2 = (suggest_T *)s2;
|
|
344
|
10602 int n = p1->st_score - p2->st_score;
|
|
|
10603
|
|
|
10604 if (n == 0)
|
|
|
10605 return p1->st_altscore - p2->st_altscore;
|
|
|
10606 return n;
|
|
323
|
10607 }
|
|
|
10608
|
|
|
10609 /*
|
|
|
10610 * Cleanup the suggestions:
|
|
|
10611 * - Sort on score.
|
|
|
10612 * - Remove words that won't be displayed.
|
|
344
|
10613 * Returns the maximum score in the list or "maxscore" unmodified.
|
|
323
|
10614 */
|
|
344
|
10615 static int
|
|
|
10616 cleanup_suggestions(gap, maxscore, keep)
|
|
|
10617 garray_T *gap;
|
|
|
10618 int maxscore;
|
|
324
|
10619 int keep; /* nr of suggestions to keep */
|
|
323
|
10620 {
|
|
344
|
10621 suggest_T *stp = &SUG(*gap, 0);
|
|
323
|
10622 int i;
|
|
|
10623
|
|
|
10624 /* Sort the list. */
|
|
344
|
10625 qsort(gap->ga_data, (size_t)gap->ga_len, sizeof(suggest_T), sug_compare);
|
|
323
|
10626
|
|
|
10627 /* Truncate the list to the number of suggestions that will be displayed. */
|
|
344
|
10628 if (gap->ga_len > keep)
|
|
323
|
10629 {
|
|
344
|
10630 for (i = keep; i < gap->ga_len; ++i)
|
|
323
|
10631 vim_free(stp[i].st_word);
|
|
344
|
10632 gap->ga_len = keep;
|
|
|
10633 return stp[keep - 1].st_score;
|
|
323
|
10634 }
|
|
344
|
10635 return maxscore;
|
|
323
|
10636 }
|
|
|
10637
|
|
372
|
10638 #if defined(FEAT_EVAL) || defined(PROTO)
|
|
|
10639 /*
|
|
|
10640 * Soundfold a string, for soundfold().
|
|
|
10641 * Result is in allocated memory, NULL for an error.
|
|
|
10642 */
|
|
|
10643 char_u *
|
|
|
10644 eval_soundfold(word)
|
|
|
10645 char_u *word;
|
|
|
10646 {
|
|
|
10647 langp_T *lp;
|
|
|
10648 char_u sound[MAXWLEN];
|
|
|
10649
|
|
|
10650 if (curwin->w_p_spell && *curbuf->b_p_spl != NUL)
|
|
|
10651 /* Use the sound-folding of the first language that supports it. */
|
|
|
10652 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
10653 lp->lp_slang != NULL; ++lp)
|
|
|
10654 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
10655 {
|
|
|
10656 /* soundfold the word */
|
|
375
|
10657 spell_soundfold(lp->lp_slang, word, FALSE, sound);
|
|
372
|
10658 return vim_strsave(sound);
|
|
|
10659 }
|
|
|
10660
|
|
|
10661 /* No language with sound folding, return word as-is. */
|
|
|
10662 return vim_strsave(word);
|
|
|
10663 }
|
|
|
10664 #endif
|
|
|
10665
|
|
323
|
10666 /*
|
|
|
10667 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]".
|
|
485
|
10668 *
|
|
|
10669 * There are many ways to turn a word into a sound-a-like representation. The
|
|
|
10670 * oldest is Soundex (1918!). A nice overview can be found in "Approximate
|
|
|
10671 * swedish name matching - survey and test of different algorithms" by Klas
|
|
|
10672 * Erikson.
|
|
|
10673 *
|
|
|
10674 * We support two methods:
|
|
|
10675 * 1. SOFOFROM/SOFOTO do a simple character mapping.
|
|
|
10676 * 2. SAL items define a more advanced sound-folding (and much slower).
|
|
323
|
10677 */
|
|
|
10678 static void
|
|
375
|
10679 spell_soundfold(slang, inword, folded, res)
|
|
|
10680 slang_T *slang;
|
|
|
10681 char_u *inword;
|
|
|
10682 int folded; /* "inword" is already case-folded */
|
|
|
10683 char_u *res;
|
|
|
10684 {
|
|
|
10685 char_u fword[MAXWLEN];
|
|
|
10686 char_u *word;
|
|
|
10687
|
|
|
10688 if (slang->sl_sofo)
|
|
|
10689 /* SOFOFROM and SOFOTO used */
|
|
|
10690 spell_soundfold_sofo(slang, inword, res);
|
|
|
10691 else
|
|
|
10692 {
|
|
|
10693 /* SAL items used. Requires the word to be case-folded. */
|
|
|
10694 if (folded)
|
|
|
10695 word = inword;
|
|
|
10696 else
|
|
|
10697 {
|
|
|
10698 (void)spell_casefold(inword, STRLEN(inword), fword, MAXWLEN);
|
|
|
10699 word = fword;
|
|
|
10700 }
|
|
|
10701
|
|
|
10702 #ifdef FEAT_MBYTE
|
|
|
10703 if (has_mbyte)
|
|
|
10704 spell_soundfold_wsal(slang, word, res);
|
|
|
10705 else
|
|
|
10706 #endif
|
|
|
10707 spell_soundfold_sal(slang, word, res);
|
|
|
10708 }
|
|
|
10709 }
|
|
|
10710
|
|
|
10711 /*
|
|
|
10712 * Perform sound folding of "inword" into "res" according to SOFOFROM and
|
|
|
10713 * SOFOTO lines.
|
|
|
10714 */
|
|
|
10715 static void
|
|
|
10716 spell_soundfold_sofo(slang, inword, res)
|
|
|
10717 slang_T *slang;
|
|
|
10718 char_u *inword;
|
|
|
10719 char_u *res;
|
|
|
10720 {
|
|
|
10721 char_u *s;
|
|
|
10722 int ri = 0;
|
|
|
10723 int c;
|
|
|
10724
|
|
|
10725 #ifdef FEAT_MBYTE
|
|
|
10726 if (has_mbyte)
|
|
|
10727 {
|
|
|
10728 int prevc = 0;
|
|
|
10729 int *ip;
|
|
|
10730
|
|
|
10731 /* The sl_sal_first[] table contains the translation for chars up to
|
|
|
10732 * 255, sl_sal the rest. */
|
|
|
10733 for (s = inword; *s != NUL; )
|
|
|
10734 {
|
|
474
|
10735 c = mb_cptr2char_adv(&s);
|
|
375
|
10736 if (enc_utf8 ? utf_class(c) == 0 : vim_iswhite(c))
|
|
|
10737 c = ' ';
|
|
|
10738 else if (c < 256)
|
|
|
10739 c = slang->sl_sal_first[c];
|
|
|
10740 else
|
|
|
10741 {
|
|
|
10742 ip = ((int **)slang->sl_sal.ga_data)[c & 0xff];
|
|
|
10743 if (ip == NULL) /* empty list, can't match */
|
|
|
10744 c = NUL;
|
|
|
10745 else
|
|
|
10746 for (;;) /* find "c" in the list */
|
|
|
10747 {
|
|
|
10748 if (*ip == 0) /* not found */
|
|
|
10749 {
|
|
|
10750 c = NUL;
|
|
|
10751 break;
|
|
|
10752 }
|
|
|
10753 if (*ip == c) /* match! */
|
|
|
10754 {
|
|
|
10755 c = ip[1];
|
|
|
10756 break;
|
|
|
10757 }
|
|
|
10758 ip += 2;
|
|
|
10759 }
|
|
|
10760 }
|
|
|
10761
|
|
|
10762 if (c != NUL && c != prevc)
|
|
|
10763 {
|
|
|
10764 ri += mb_char2bytes(c, res + ri);
|
|
|
10765 if (ri + MB_MAXBYTES > MAXWLEN)
|
|
|
10766 break;
|
|
|
10767 prevc = c;
|
|
|
10768 }
|
|
|
10769 }
|
|
|
10770 }
|
|
|
10771 else
|
|
|
10772 #endif
|
|
|
10773 {
|
|
|
10774 /* The sl_sal_first[] table contains the translation. */
|
|
|
10775 for (s = inword; (c = *s) != NUL; ++s)
|
|
|
10776 {
|
|
|
10777 if (vim_iswhite(c))
|
|
|
10778 c = ' ';
|
|
|
10779 else
|
|
|
10780 c = slang->sl_sal_first[c];
|
|
|
10781 if (c != NUL && (ri == 0 || res[ri - 1] != c))
|
|
|
10782 res[ri++] = c;
|
|
|
10783 }
|
|
|
10784 }
|
|
|
10785
|
|
|
10786 res[ri] = NUL;
|
|
|
10787 }
|
|
|
10788
|
|
|
10789 static void
|
|
|
10790 spell_soundfold_sal(slang, inword, res)
|
|
323
|
10791 slang_T *slang;
|
|
|
10792 char_u *inword;
|
|
|
10793 char_u *res;
|
|
|
10794 {
|
|
344
|
10795 salitem_T *smp;
|
|
323
|
10796 char_u word[MAXWLEN];
|
|
375
|
10797 char_u *s = inword;
|
|
323
|
10798 char_u *t;
|
|
344
|
10799 char_u *pf;
|
|
323
|
10800 int i, j, z;
|
|
344
|
10801 int reslen;
|
|
323
|
10802 int n, k = 0;
|
|
|
10803 int z0;
|
|
|
10804 int k0;
|
|
|
10805 int n0;
|
|
|
10806 int c;
|
|
|
10807 int pri;
|
|
|
10808 int p0 = -333;
|
|
|
10809 int c0;
|
|
|
10810
|
|
324
|
10811 /* Remove accents, if wanted. We actually remove all non-word characters.
|
|
375
|
10812 * But keep white space. We need a copy, the word may be changed here. */
|
|
323
|
10813 if (slang->sl_rem_accents)
|
|
|
10814 {
|
|
|
10815 t = word;
|
|
375
|
10816 while (*s != NUL)
|
|
323
|
10817 {
|
|
324
|
10818 if (vim_iswhite(*s))
|
|
344
|
10819 {
|
|
|
10820 *t++ = ' ';
|
|
|
10821 s = skipwhite(s);
|
|
|
10822 }
|
|
323
|
10823 else
|
|
|
10824 {
|
|
376
|
10825 if (spell_iswordp_nmw(s))
|
|
323
|
10826 *t++ = *s;
|
|
|
10827 ++s;
|
|
|
10828 }
|
|
|
10829 }
|
|
|
10830 *t = NUL;
|
|
|
10831 }
|
|
|
10832 else
|
|
375
|
10833 STRCPY(word, s);
|
|
323
|
10834
|
|
344
|
10835 smp = (salitem_T *)slang->sl_sal.ga_data;
|
|
323
|
10836
|
|
|
10837 /*
|
|
|
10838 * This comes from Aspell phonet.cpp. Converted from C++ to C.
|
|
324
|
10839 * Changed to keep spaces.
|
|
323
|
10840 */
|
|
344
|
10841 i = reslen = z = 0;
|
|
323
|
10842 while ((c = word[i]) != NUL)
|
|
|
10843 {
|
|
344
|
10844 /* Start with the first rule that has the character in the word. */
|
|
323
|
10845 n = slang->sl_sal_first[c];
|
|
|
10846 z0 = 0;
|
|
|
10847
|
|
|
10848 if (n >= 0)
|
|
|
10849 {
|
|
|
10850 /* check all rules for the same letter */
|
|
344
|
10851 for (; (s = smp[n].sm_lead)[0] == c; ++n)
|
|
323
|
10852 {
|
|
344
|
10853 /* Quickly skip entries that don't match the word. Most
|
|
|
10854 * entries are less then three chars, optimize for that. */
|
|
|
10855 k = smp[n].sm_leadlen;
|
|
|
10856 if (k > 1)
|
|
323
|
10857 {
|
|
344
|
10858 if (word[i + 1] != s[1])
|
|
|
10859 continue;
|
|
|
10860 if (k > 2)
|
|
|
10861 {
|
|
|
10862 for (j = 2; j < k; ++j)
|
|
|
10863 if (word[i + j] != s[j])
|
|
|
10864 break;
|
|
|
10865 if (j < k)
|
|
|
10866 continue;
|
|
|
10867 }
|
|
323
|
10868 }
|
|
|
10869
|
|
375
|
10870 if ((pf = smp[n].sm_oneof) != NULL)
|
|
323
|
10871 {
|
|
375
|
10872 /* Check for match with one of the chars in "sm_oneof". */
|
|
344
|
10873 while (*pf != NUL && *pf != word[i + k])
|
|
|
10874 ++pf;
|
|
|
10875 if (*pf == NUL)
|
|
|
10876 continue;
|
|
|
10877 ++k;
|
|
323
|
10878 }
|
|
344
|
10879 s = smp[n].sm_rules;
|
|
|
10880 pri = 5; /* default priority */
|
|
323
|
10881
|
|
|
10882 p0 = *s;
|
|
|
10883 k0 = k;
|
|
|
10884 while (*s == '-' && k > 1)
|
|
|
10885 {
|
|
|
10886 k--;
|
|
|
10887 s++;
|
|
|
10888 }
|
|
|
10889 if (*s == '<')
|
|
|
10890 s++;
|
|
344
|
10891 if (VIM_ISDIGIT(*s))
|
|
323
|
10892 {
|
|
|
10893 /* determine priority */
|
|
|
10894 pri = *s - '0';
|
|
|
10895 s++;
|
|
|
10896 }
|
|
|
10897 if (*s == '^' && *(s + 1) == '^')
|
|
|
10898 s++;
|
|
|
10899
|
|
|
10900 if (*s == NUL
|
|
|
10901 || (*s == '^'
|
|
324
|
10902 && (i == 0 || !(word[i - 1] == ' '
|
|
376
|
10903 || spell_iswordp(word + i - 1, curbuf)))
|
|
323
|
10904 && (*(s + 1) != '$'
|
|
376
|
10905 || (!spell_iswordp(word + i + k0, curbuf))))
|
|
323
|
10906 || (*s == '$' && i > 0
|
|
376
|
10907 && spell_iswordp(word + i - 1, curbuf)
|
|
|
10908 && (!spell_iswordp(word + i + k0, curbuf))))
|
|
323
|
10909 {
|
|
|
10910 /* search for followup rules, if: */
|
|
|
10911 /* followup and k > 1 and NO '-' in searchstring */
|
|
|
10912 c0 = word[i + k - 1];
|
|
|
10913 n0 = slang->sl_sal_first[c0];
|
|
|
10914
|
|
|
10915 if (slang->sl_followup && k > 1 && n0 >= 0
|
|
344
|
10916 && p0 != '-' && word[i + k] != NUL)
|
|
323
|
10917 {
|
|
|
10918 /* test follow-up rule for "word[i + k]" */
|
|
344
|
10919 for ( ; (s = smp[n0].sm_lead)[0] == c0; ++n0)
|
|
323
|
10920 {
|
|
344
|
10921 /* Quickly skip entries that don't match the word.
|
|
|
10922 * */
|
|
|
10923 k0 = smp[n0].sm_leadlen;
|
|
|
10924 if (k0 > 1)
|
|
323
|
10925 {
|
|
344
|
10926 if (word[i + k] != s[1])
|
|
|
10927 continue;
|
|
|
10928 if (k0 > 2)
|
|
|
10929 {
|
|
|
10930 pf = word + i + k + 1;
|
|
|
10931 for (j = 2; j < k0; ++j)
|
|
|
10932 if (*pf++ != s[j])
|
|
|
10933 break;
|
|
|
10934 if (j < k0)
|
|
|
10935 continue;
|
|
|
10936 }
|
|
323
|
10937 }
|
|
344
|
10938 k0 += k - 1;
|
|
|
10939
|
|
375
|
10940 if ((pf = smp[n0].sm_oneof) != NULL)
|
|
323
|
10941 {
|
|
344
|
10942 /* Check for match with one of the chars in
|
|
375
|
10943 * "sm_oneof". */
|
|
344
|
10944 while (*pf != NUL && *pf != word[i + k0])
|
|
|
10945 ++pf;
|
|
|
10946 if (*pf == NUL)
|
|
|
10947 continue;
|
|
|
10948 ++k0;
|
|
323
|
10949 }
|
|
344
|
10950
|
|
|
10951 p0 = 5;
|
|
|
10952 s = smp[n0].sm_rules;
|
|
323
|
10953 while (*s == '-')
|
|
|
10954 {
|
|
344
|
10955 /* "k0" gets NOT reduced because
|
|
|
10956 * "if (k0 == k)" */
|
|
323
|
10957 s++;
|
|
|
10958 }
|
|
|
10959 if (*s == '<')
|
|
|
10960 s++;
|
|
344
|
10961 if (VIM_ISDIGIT(*s))
|
|
323
|
10962 {
|
|
|
10963 p0 = *s - '0';
|
|
|
10964 s++;
|
|
|
10965 }
|
|
|
10966
|
|
|
10967 if (*s == NUL
|
|
|
10968 /* *s == '^' cuts */
|
|
|
10969 || (*s == '$'
|
|
376
|
10970 && !spell_iswordp(word + i + k0,
|
|
|
10971 curbuf)))
|
|
323
|
10972 {
|
|
|
10973 if (k0 == k)
|
|
|
10974 /* this is just a piece of the string */
|
|
|
10975 continue;
|
|
|
10976
|
|
|
10977 if (p0 < pri)
|
|
|
10978 /* priority too low */
|
|
|
10979 continue;
|
|
|
10980 /* rule fits; stop search */
|
|
|
10981 break;
|
|
|
10982 }
|
|
|
10983 }
|
|
|
10984
|
|
344
|
10985 if (p0 >= pri && smp[n0].sm_lead[0] == c0)
|
|
323
|
10986 continue;
|
|
|
10987 }
|
|
|
10988
|
|
|
10989 /* replace string */
|
|
344
|
10990 s = smp[n].sm_to;
|
|
389
|
10991 if (s == NULL)
|
|
|
10992 s = (char_u *)"";
|
|
344
|
10993 pf = smp[n].sm_rules;
|
|
|
10994 p0 = (vim_strchr(pf, '<') != NULL) ? 1 : 0;
|
|
323
|
10995 if (p0 == 1 && z == 0)
|
|
|
10996 {
|
|
|
10997 /* rule with '<' is used */
|
|
344
|
10998 if (reslen > 0 && *s != NUL && (res[reslen - 1] == c
|
|
|
10999 || res[reslen - 1] == *s))
|
|
|
11000 reslen--;
|
|
323
|
11001 z0 = 1;
|
|
|
11002 z = 1;
|
|
|
11003 k0 = 0;
|
|
372
|
11004 while (*s != NUL && word[i + k0] != NUL)
|
|
323
|
11005 {
|
|
|
11006 word[i + k0] = *s;
|
|
|
11007 k0++;
|
|
|
11008 s++;
|
|
|
11009 }
|
|
|
11010 if (k > k0)
|
|
|
11011 mch_memmove(word + i + k0, word + i + k,
|
|
|
11012 STRLEN(word + i + k) + 1);
|
|
|
11013
|
|
|
11014 /* new "actual letter" */
|
|
|
11015 c = word[i];
|
|
|
11016 }
|
|
|
11017 else
|
|
|
11018 {
|
|
|
11019 /* no '<' rule used */
|
|
|
11020 i += k - 1;
|
|
|
11021 z = 0;
|
|
344
|
11022 while (*s != NUL && s[1] != NUL && reslen < MAXWLEN)
|
|
323
|
11023 {
|
|
344
|
11024 if (reslen == 0 || res[reslen - 1] != *s)
|
|
372
|
11025 res[reslen++] = *s;
|
|
323
|
11026 s++;
|
|
|
11027 }
|
|
|
11028 /* new "actual letter" */
|
|
|
11029 c = *s;
|
|
344
|
11030 if (strstr((char *)pf, "^^") != NULL)
|
|
323
|
11031 {
|
|
|
11032 if (c != NUL)
|
|
372
|
11033 res[reslen++] = c;
|
|
323
|
11034 mch_memmove(word, word + i + 1,
|
|
|
11035 STRLEN(word + i + 1) + 1);
|
|
|
11036 i = 0;
|
|
|
11037 z0 = 1;
|
|
|
11038 }
|
|
|
11039 }
|
|
|
11040 break;
|
|
|
11041 }
|
|
|
11042 }
|
|
|
11043 }
|
|
324
|
11044 else if (vim_iswhite(c))
|
|
|
11045 {
|
|
|
11046 c = ' ';
|
|
|
11047 k = 1;
|
|
|
11048 }
|
|
323
|
11049
|
|
|
11050 if (z0 == 0)
|
|
|
11051 {
|
|
344
|
11052 if (k && !p0 && reslen < MAXWLEN && c != NUL
|
|
|
11053 && (!slang->sl_collapse || reslen == 0
|
|
|
11054 || res[reslen - 1] != c))
|
|
323
|
11055 /* condense only double letters */
|
|
372
|
11056 res[reslen++] = c;
|
|
323
|
11057
|
|
|
11058 i++;
|
|
|
11059 z = 0;
|
|
|
11060 k = 0;
|
|
|
11061 }
|
|
|
11062 }
|
|
|
11063
|
|
344
|
11064 res[reslen] = NUL;
|
|
323
|
11065 }
|
|
|
11066
|
|
372
|
11067 #ifdef FEAT_MBYTE
|
|
|
11068 /*
|
|
|
11069 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]".
|
|
|
11070 * Multi-byte version of spell_soundfold().
|
|
|
11071 */
|
|
|
11072 static void
|
|
375
|
11073 spell_soundfold_wsal(slang, inword, res)
|
|
372
|
11074 slang_T *slang;
|
|
|
11075 char_u *inword;
|
|
|
11076 char_u *res;
|
|
|
11077 {
|
|
375
|
11078 salitem_T *smp = (salitem_T *)slang->sl_sal.ga_data;
|
|
372
|
11079 int word[MAXWLEN];
|
|
|
11080 int wres[MAXWLEN];
|
|
|
11081 int l;
|
|
|
11082 char_u *s;
|
|
|
11083 int *ws;
|
|
|
11084 char_u *t;
|
|
|
11085 int *pf;
|
|
|
11086 int i, j, z;
|
|
|
11087 int reslen;
|
|
|
11088 int n, k = 0;
|
|
|
11089 int z0;
|
|
|
11090 int k0;
|
|
|
11091 int n0;
|
|
|
11092 int c;
|
|
|
11093 int pri;
|
|
|
11094 int p0 = -333;
|
|
|
11095 int c0;
|
|
|
11096 int did_white = FALSE;
|
|
|
11097
|
|
|
11098 /*
|
|
|
11099 * Convert the multi-byte string to a wide-character string.
|
|
|
11100 * Remove accents, if wanted. We actually remove all non-word characters.
|
|
|
11101 * But keep white space.
|
|
|
11102 */
|
|
|
11103 n = 0;
|
|
|
11104 for (s = inword; *s != NUL; )
|
|
|
11105 {
|
|
|
11106 t = s;
|
|
474
|
11107 c = mb_cptr2char_adv(&s);
|
|
372
|
11108 if (slang->sl_rem_accents)
|
|
|
11109 {
|
|
|
11110 if (enc_utf8 ? utf_class(c) == 0 : vim_iswhite(c))
|
|
|
11111 {
|
|
|
11112 if (did_white)
|
|
|
11113 continue;
|
|
|
11114 c = ' ';
|
|
|
11115 did_white = TRUE;
|
|
|
11116 }
|
|
|
11117 else
|
|
|
11118 {
|
|
|
11119 did_white = FALSE;
|
|
376
|
11120 if (!spell_iswordp_nmw(t))
|
|
372
|
11121 continue;
|
|
|
11122 }
|
|
|
11123 }
|
|
|
11124 word[n++] = c;
|
|
|
11125 }
|
|
|
11126 word[n] = NUL;
|
|
|
11127
|
|
|
11128 /*
|
|
|
11129 * This comes from Aspell phonet.cpp.
|
|
|
11130 * Converted from C++ to C. Added support for multi-byte chars.
|
|
|
11131 * Changed to keep spaces.
|
|
|
11132 */
|
|
|
11133 i = reslen = z = 0;
|
|
|
11134 while ((c = word[i]) != NUL)
|
|
|
11135 {
|
|
|
11136 /* Start with the first rule that has the character in the word. */
|
|
|
11137 n = slang->sl_sal_first[c & 0xff];
|
|
|
11138 z0 = 0;
|
|
|
11139
|
|
|
11140 if (n >= 0)
|
|
|
11141 {
|
|
375
|
11142 /* check all rules for the same index byte */
|
|
372
|
11143 for (; ((ws = smp[n].sm_lead_w)[0] & 0xff) == (c & 0xff); ++n)
|
|
|
11144 {
|
|
|
11145 /* Quickly skip entries that don't match the word. Most
|
|
|
11146 * entries are less then three chars, optimize for that. */
|
|
375
|
11147 if (c != ws[0])
|
|
|
11148 continue;
|
|
372
|
11149 k = smp[n].sm_leadlen;
|
|
|
11150 if (k > 1)
|
|
|
11151 {
|
|
|
11152 if (word[i + 1] != ws[1])
|
|
|
11153 continue;
|
|
|
11154 if (k > 2)
|
|
|
11155 {
|
|
|
11156 for (j = 2; j < k; ++j)
|
|
|
11157 if (word[i + j] != ws[j])
|
|
|
11158 break;
|
|
|
11159 if (j < k)
|
|
|
11160 continue;
|
|
|
11161 }
|
|
|
11162 }
|
|
|
11163
|
|
375
|
11164 if ((pf = smp[n].sm_oneof_w) != NULL)
|
|
372
|
11165 {
|
|
375
|
11166 /* Check for match with one of the chars in "sm_oneof". */
|
|
372
|
11167 while (*pf != NUL && *pf != word[i + k])
|
|
|
11168 ++pf;
|
|
|
11169 if (*pf == NUL)
|
|
|
11170 continue;
|
|
|
11171 ++k;
|
|
|
11172 }
|
|
|
11173 s = smp[n].sm_rules;
|
|
|
11174 pri = 5; /* default priority */
|
|
|
11175
|
|
|
11176 p0 = *s;
|
|
|
11177 k0 = k;
|
|
|
11178 while (*s == '-' && k > 1)
|
|
|
11179 {
|
|
|
11180 k--;
|
|
|
11181 s++;
|
|
|
11182 }
|
|
|
11183 if (*s == '<')
|
|
|
11184 s++;
|
|
|
11185 if (VIM_ISDIGIT(*s))
|
|
|
11186 {
|
|
|
11187 /* determine priority */
|
|
|
11188 pri = *s - '0';
|
|
|
11189 s++;
|
|
|
11190 }
|
|
|
11191 if (*s == '^' && *(s + 1) == '^')
|
|
|
11192 s++;
|
|
|
11193
|
|
|
11194 if (*s == NUL
|
|
|
11195 || (*s == '^'
|
|
|
11196 && (i == 0 || !(word[i - 1] == ' '
|
|
376
|
11197 || spell_iswordp_w(word + i - 1, curbuf)))
|
|
372
|
11198 && (*(s + 1) != '$'
|
|
376
|
11199 || (!spell_iswordp_w(word + i + k0, curbuf))))
|
|
372
|
11200 || (*s == '$' && i > 0
|
|
376
|
11201 && spell_iswordp_w(word + i - 1, curbuf)
|
|
|
11202 && (!spell_iswordp_w(word + i + k0, curbuf))))
|
|
372
|
11203 {
|
|
|
11204 /* search for followup rules, if: */
|
|
|
11205 /* followup and k > 1 and NO '-' in searchstring */
|
|
|
11206 c0 = word[i + k - 1];
|
|
|
11207 n0 = slang->sl_sal_first[c0 & 0xff];
|
|
|
11208
|
|
|
11209 if (slang->sl_followup && k > 1 && n0 >= 0
|
|
|
11210 && p0 != '-' && word[i + k] != NUL)
|
|
|
11211 {
|
|
375
|
11212 /* Test follow-up rule for "word[i + k]"; loop over
|
|
|
11213 * all entries with the same index byte. */
|
|
372
|
11214 for ( ; ((ws = smp[n0].sm_lead_w)[0] & 0xff)
|
|
|
11215 == (c0 & 0xff); ++n0)
|
|
|
11216 {
|
|
|
11217 /* Quickly skip entries that don't match the word.
|
|
375
|
11218 */
|
|
|
11219 if (c0 != ws[0])
|
|
|
11220 continue;
|
|
372
|
11221 k0 = smp[n0].sm_leadlen;
|
|
|
11222 if (k0 > 1)
|
|
|
11223 {
|
|
|
11224 if (word[i + k] != ws[1])
|
|
|
11225 continue;
|
|
|
11226 if (k0 > 2)
|
|
|
11227 {
|
|
|
11228 pf = word + i + k + 1;
|
|
|
11229 for (j = 2; j < k0; ++j)
|
|
|
11230 if (*pf++ != ws[j])
|
|
|
11231 break;
|
|
|
11232 if (j < k0)
|
|
|
11233 continue;
|
|
|
11234 }
|
|
|
11235 }
|
|
|
11236 k0 += k - 1;
|
|
|
11237
|
|
375
|
11238 if ((pf = smp[n0].sm_oneof_w) != NULL)
|
|
372
|
11239 {
|
|
|
11240 /* Check for match with one of the chars in
|
|
375
|
11241 * "sm_oneof". */
|
|
372
|
11242 while (*pf != NUL && *pf != word[i + k0])
|
|
|
11243 ++pf;
|
|
|
11244 if (*pf == NUL)
|
|
|
11245 continue;
|
|
|
11246 ++k0;
|
|
|
11247 }
|
|
|
11248
|
|
|
11249 p0 = 5;
|
|
|
11250 s = smp[n0].sm_rules;
|
|
|
11251 while (*s == '-')
|
|
|
11252 {
|
|
|
11253 /* "k0" gets NOT reduced because
|
|
|
11254 * "if (k0 == k)" */
|
|
|
11255 s++;
|
|
|
11256 }
|
|
|
11257 if (*s == '<')
|
|
|
11258 s++;
|
|
|
11259 if (VIM_ISDIGIT(*s))
|
|
|
11260 {
|
|
|
11261 p0 = *s - '0';
|
|
|
11262 s++;
|
|
|
11263 }
|
|
|
11264
|
|
|
11265 if (*s == NUL
|
|
|
11266 /* *s == '^' cuts */
|
|
|
11267 || (*s == '$'
|
|
376
|
11268 && !spell_iswordp_w(word + i + k0,
|
|
|
11269 curbuf)))
|
|
372
|
11270 {
|
|
|
11271 if (k0 == k)
|
|
|
11272 /* this is just a piece of the string */
|
|
|
11273 continue;
|
|
|
11274
|
|
|
11275 if (p0 < pri)
|
|
|
11276 /* priority too low */
|
|
|
11277 continue;
|
|
|
11278 /* rule fits; stop search */
|
|
|
11279 break;
|
|
|
11280 }
|
|
|
11281 }
|
|
|
11282
|
|
|
11283 if (p0 >= pri && (smp[n0].sm_lead_w[0] & 0xff)
|
|
|
11284 == (c0 & 0xff))
|
|
|
11285 continue;
|
|
|
11286 }
|
|
|
11287
|
|
|
11288 /* replace string */
|
|
|
11289 ws = smp[n].sm_to_w;
|
|
|
11290 s = smp[n].sm_rules;
|
|
|
11291 p0 = (vim_strchr(s, '<') != NULL) ? 1 : 0;
|
|
|
11292 if (p0 == 1 && z == 0)
|
|
|
11293 {
|
|
|
11294 /* rule with '<' is used */
|
|
389
|
11295 if (reslen > 0 && ws != NULL && *ws != NUL
|
|
|
11296 && (wres[reslen - 1] == c
|
|
372
|
11297 || wres[reslen - 1] == *ws))
|
|
|
11298 reslen--;
|
|
|
11299 z0 = 1;
|
|
|
11300 z = 1;
|
|
|
11301 k0 = 0;
|
|
389
|
11302 if (ws != NULL)
|
|
|
11303 while (*ws != NUL && word[i + k0] != NUL)
|
|
|
11304 {
|
|
|
11305 word[i + k0] = *ws;
|
|
|
11306 k0++;
|
|
|
11307 ws++;
|
|
|
11308 }
|
|
372
|
11309 if (k > k0)
|
|
|
11310 mch_memmove(word + i + k0, word + i + k,
|
|
|
11311 sizeof(int) * (STRLEN(word + i + k) + 1));
|
|
|
11312
|
|
|
11313 /* new "actual letter" */
|
|
|
11314 c = word[i];
|
|
|
11315 }
|
|
|
11316 else
|
|
|
11317 {
|
|
|
11318 /* no '<' rule used */
|
|
|
11319 i += k - 1;
|
|
|
11320 z = 0;
|
|
389
|
11321 if (ws != NULL)
|
|
|
11322 while (*ws != NUL && ws[1] != NUL
|
|
|
11323 && reslen < MAXWLEN)
|
|
|
11324 {
|
|
|
11325 if (reslen == 0 || wres[reslen - 1] != *ws)
|
|
|
11326 wres[reslen++] = *ws;
|
|
|
11327 ws++;
|
|
|
11328 }
|
|
372
|
11329 /* new "actual letter" */
|
|
389
|
11330 if (ws == NULL)
|
|
|
11331 c = NUL;
|
|
|
11332 else
|
|
|
11333 c = *ws;
|
|
372
|
11334 if (strstr((char *)s, "^^") != NULL)
|
|
|
11335 {
|
|
|
11336 if (c != NUL)
|
|
|
11337 wres[reslen++] = c;
|
|
|
11338 mch_memmove(word, word + i + 1,
|
|
|
11339 sizeof(int) * (STRLEN(word + i + 1) + 1));
|
|
|
11340 i = 0;
|
|
|
11341 z0 = 1;
|
|
|
11342 }
|
|
|
11343 }
|
|
|
11344 break;
|
|
|
11345 }
|
|
|
11346 }
|
|
|
11347 }
|
|
|
11348 else if (vim_iswhite(c))
|
|
|
11349 {
|
|
|
11350 c = ' ';
|
|
|
11351 k = 1;
|
|
|
11352 }
|
|
|
11353
|
|
|
11354 if (z0 == 0)
|
|
|
11355 {
|
|
|
11356 if (k && !p0 && reslen < MAXWLEN && c != NUL
|
|
|
11357 && (!slang->sl_collapse || reslen == 0
|
|
|
11358 || wres[reslen - 1] != c))
|
|
|
11359 /* condense only double letters */
|
|
|
11360 wres[reslen++] = c;
|
|
|
11361
|
|
|
11362 i++;
|
|
|
11363 z = 0;
|
|
|
11364 k = 0;
|
|
|
11365 }
|
|
|
11366 }
|
|
|
11367
|
|
|
11368 /* Convert wide characters in "wres" to a multi-byte string in "res". */
|
|
|
11369 l = 0;
|
|
|
11370 for (n = 0; n < reslen; ++n)
|
|
|
11371 {
|
|
|
11372 l += mb_char2bytes(wres[n], res + l);
|
|
|
11373 if (l + MB_MAXBYTES > MAXWLEN)
|
|
|
11374 break;
|
|
|
11375 }
|
|
|
11376 res[l] = NUL;
|
|
|
11377 }
|
|
|
11378 #endif
|
|
|
11379
|
|
324
|
11380 /*
|
|
344
|
11381 * Compute a score for two sound-a-like words.
|
|
|
11382 * This permits up to two inserts/deletes/swaps/etc. to keep things fast.
|
|
|
11383 * Instead of a generic loop we write out the code. That keeps it fast by
|
|
|
11384 * avoiding checks that will not be possible.
|
|
|
11385 */
|
|
|
11386 static int
|
|
351
|
11387 soundalike_score(goodstart, badstart)
|
|
|
11388 char_u *goodstart; /* sound-folded good word */
|
|
|
11389 char_u *badstart; /* sound-folded bad word */
|
|
344
|
11390 {
|
|
351
|
11391 char_u *goodsound = goodstart;
|
|
|
11392 char_u *badsound = badstart;
|
|
|
11393 int goodlen;
|
|
|
11394 int badlen;
|
|
344
|
11395 int n;
|
|
|
11396 char_u *pl, *ps;
|
|
|
11397 char_u *pl2, *ps2;
|
|
351
|
11398 int score = 0;
|
|
|
11399
|
|
|
11400 /* adding/inserting "*" at the start (word starts with vowel) shouldn't be
|
|
|
11401 * counted so much, vowels halfway the word aren't counted at all. */
|
|
|
11402 if ((*badsound == '*' || *goodsound == '*') && *badsound != *goodsound)
|
|
|
11403 {
|
|
|
11404 score = SCORE_DEL / 2;
|
|
|
11405 if (*badsound == '*')
|
|
|
11406 ++badsound;
|
|
|
11407 else
|
|
|
11408 ++goodsound;
|
|
|
11409 }
|
|
|
11410
|
|
|
11411 goodlen = STRLEN(goodsound);
|
|
|
11412 badlen = STRLEN(badsound);
|
|
344
|
11413
|
|
|
11414 /* Return quickly if the lenghts are too different to be fixed by two
|
|
|
11415 * changes. */
|
|
|
11416 n = goodlen - badlen;
|
|
|
11417 if (n < -2 || n > 2)
|
|
|
11418 return SCORE_MAXMAX;
|
|
|
11419
|
|
|
11420 if (n > 0)
|
|
|
11421 {
|
|
351
|
11422 pl = goodsound; /* goodsound is longest */
|
|
344
|
11423 ps = badsound;
|
|
|
11424 }
|
|
|
11425 else
|
|
|
11426 {
|
|
351
|
11427 pl = badsound; /* badsound is longest */
|
|
344
|
11428 ps = goodsound;
|
|
|
11429 }
|
|
|
11430
|
|
|
11431 /* Skip over the identical part. */
|
|
|
11432 while (*pl == *ps && *pl != NUL)
|
|
|
11433 {
|
|
|
11434 ++pl;
|
|
|
11435 ++ps;
|
|
|
11436 }
|
|
|
11437
|
|
|
11438 switch (n)
|
|
|
11439 {
|
|
|
11440 case -2:
|
|
|
11441 case 2:
|
|
|
11442 /*
|
|
|
11443 * Must delete two characters from "pl".
|
|
|
11444 */
|
|
|
11445 ++pl; /* first delete */
|
|
|
11446 while (*pl == *ps)
|
|
|
11447 {
|
|
|
11448 ++pl;
|
|
|
11449 ++ps;
|
|
|
11450 }
|
|
|
11451 /* strings must be equal after second delete */
|
|
|
11452 if (STRCMP(pl + 1, ps) == 0)
|
|
351
|
11453 return score + SCORE_DEL * 2;
|
|
344
|
11454
|
|
|
11455 /* Failed to compare. */
|
|
|
11456 break;
|
|
|
11457
|
|
|
11458 case -1:
|
|
|
11459 case 1:
|
|
|
11460 /*
|
|
|
11461 * Minimal one delete from "pl" required.
|
|
|
11462 */
|
|
|
11463
|
|
|
11464 /* 1: delete */
|
|
|
11465 pl2 = pl + 1;
|
|
|
11466 ps2 = ps;
|
|
|
11467 while (*pl2 == *ps2)
|
|
|
11468 {
|
|
|
11469 if (*pl2 == NUL) /* reached the end */
|
|
351
|
11470 return score + SCORE_DEL;
|
|
344
|
11471 ++pl2;
|
|
|
11472 ++ps2;
|
|
|
11473 }
|
|
|
11474
|
|
|
11475 /* 2: delete then swap, then rest must be equal */
|
|
|
11476 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
11477 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
11478 return score + SCORE_DEL + SCORE_SWAP;
|
|
344
|
11479
|
|
|
11480 /* 3: delete then substitute, then the rest must be equal */
|
|
|
11481 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
11482 return score + SCORE_DEL + SCORE_SUBST;
|
|
344
|
11483
|
|
|
11484 /* 4: first swap then delete */
|
|
|
11485 if (pl[0] == ps[1] && pl[1] == ps[0])
|
|
|
11486 {
|
|
|
11487 pl2 = pl + 2; /* swap, skip two chars */
|
|
|
11488 ps2 = ps + 2;
|
|
|
11489 while (*pl2 == *ps2)
|
|
|
11490 {
|
|
|
11491 ++pl2;
|
|
|
11492 ++ps2;
|
|
|
11493 }
|
|
|
11494 /* delete a char and then strings must be equal */
|
|
|
11495 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
11496 return score + SCORE_SWAP + SCORE_DEL;
|
|
344
|
11497 }
|
|
|
11498
|
|
|
11499 /* 5: first substitute then delete */
|
|
|
11500 pl2 = pl + 1; /* substitute, skip one char */
|
|
|
11501 ps2 = ps + 1;
|
|
|
11502 while (*pl2 == *ps2)
|
|
|
11503 {
|
|
|
11504 ++pl2;
|
|
|
11505 ++ps2;
|
|
|
11506 }
|
|
|
11507 /* delete a char and then strings must be equal */
|
|
|
11508 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
11509 return score + SCORE_SUBST + SCORE_DEL;
|
|
344
|
11510
|
|
|
11511 /* Failed to compare. */
|
|
|
11512 break;
|
|
|
11513
|
|
|
11514 case 0:
|
|
|
11515 /*
|
|
|
11516 * Lenghts are equal, thus changes must result in same length: An
|
|
|
11517 * insert is only possible in combination with a delete.
|
|
|
11518 * 1: check if for identical strings
|
|
|
11519 */
|
|
|
11520 if (*pl == NUL)
|
|
351
|
11521 return score;
|
|
344
|
11522
|
|
|
11523 /* 2: swap */
|
|
|
11524 if (pl[0] == ps[1] && pl[1] == ps[0])
|
|
|
11525 {
|
|
|
11526 pl2 = pl + 2; /* swap, skip two chars */
|
|
|
11527 ps2 = ps + 2;
|
|
|
11528 while (*pl2 == *ps2)
|
|
|
11529 {
|
|
|
11530 if (*pl2 == NUL) /* reached the end */
|
|
351
|
11531 return score + SCORE_SWAP;
|
|
344
|
11532 ++pl2;
|
|
|
11533 ++ps2;
|
|
|
11534 }
|
|
|
11535 /* 3: swap and swap again */
|
|
|
11536 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
11537 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
11538 return score + SCORE_SWAP + SCORE_SWAP;
|
|
344
|
11539
|
|
|
11540 /* 4: swap and substitute */
|
|
|
11541 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
11542 return score + SCORE_SWAP + SCORE_SUBST;
|
|
344
|
11543 }
|
|
|
11544
|
|
|
11545 /* 5: substitute */
|
|
|
11546 pl2 = pl + 1;
|
|
|
11547 ps2 = ps + 1;
|
|
|
11548 while (*pl2 == *ps2)
|
|
|
11549 {
|
|
|
11550 if (*pl2 == NUL) /* reached the end */
|
|
351
|
11551 return score + SCORE_SUBST;
|
|
344
|
11552 ++pl2;
|
|
|
11553 ++ps2;
|
|
|
11554 }
|
|
|
11555
|
|
|
11556 /* 6: substitute and swap */
|
|
|
11557 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
11558 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
11559 return score + SCORE_SUBST + SCORE_SWAP;
|
|
344
|
11560
|
|
|
11561 /* 7: substitute and substitute */
|
|
|
11562 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
11563 return score + SCORE_SUBST + SCORE_SUBST;
|
|
344
|
11564
|
|
|
11565 /* 8: insert then delete */
|
|
|
11566 pl2 = pl;
|
|
|
11567 ps2 = ps + 1;
|
|
|
11568 while (*pl2 == *ps2)
|
|
|
11569 {
|
|
|
11570 ++pl2;
|
|
|
11571 ++ps2;
|
|
|
11572 }
|
|
|
11573 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
11574 return score + SCORE_INS + SCORE_DEL;
|
|
344
|
11575
|
|
|
11576 /* 9: delete then insert */
|
|
|
11577 pl2 = pl + 1;
|
|
|
11578 ps2 = ps;
|
|
|
11579 while (*pl2 == *ps2)
|
|
|
11580 {
|
|
|
11581 ++pl2;
|
|
|
11582 ++ps2;
|
|
|
11583 }
|
|
|
11584 if (STRCMP(pl2, ps2 + 1) == 0)
|
|
351
|
11585 return score + SCORE_INS + SCORE_DEL;
|
|
344
|
11586
|
|
|
11587 /* Failed to compare. */
|
|
|
11588 break;
|
|
|
11589 }
|
|
|
11590
|
|
|
11591 return SCORE_MAXMAX;
|
|
|
11592 }
|
|
324
|
11593
|
|
323
|
11594 /*
|
|
|
11595 * Compute the "edit distance" to turn "badword" into "goodword". The less
|
|
344
|
11596 * deletes/inserts/substitutes/swaps are required the lower the score.
|
|
324
|
11597 *
|
|
485
|
11598 * The algorithm is described by Du and Chang, 1992.
|
|
|
11599 * The implementation of the algorithm comes from Aspell editdist.cpp,
|
|
|
11600 * edit_distance(). It has been converted from C++ to C and modified to
|
|
|
11601 * support multi-byte characters.
|
|
323
|
11602 */
|
|
|
11603 static int
|
|
|
11604 spell_edit_score(badword, goodword)
|
|
|
11605 char_u *badword;
|
|
|
11606 char_u *goodword;
|
|
|
11607 {
|
|
|
11608 int *cnt;
|
|
372
|
11609 int badlen, goodlen; /* lenghts including NUL */
|
|
323
|
11610 int j, i;
|
|
|
11611 int t;
|
|
|
11612 int bc, gc;
|
|
324
|
11613 int pbc, pgc;
|
|
|
11614 #ifdef FEAT_MBYTE
|
|
|
11615 char_u *p;
|
|
|
11616 int wbadword[MAXWLEN];
|
|
|
11617 int wgoodword[MAXWLEN];
|
|
|
11618
|
|
|
11619 if (has_mbyte)
|
|
|
11620 {
|
|
|
11621 /* Get the characters from the multi-byte strings and put them in an
|
|
|
11622 * int array for easy access. */
|
|
|
11623 for (p = badword, badlen = 0; *p != NUL; )
|
|
474
|
11624 wbadword[badlen++] = mb_cptr2char_adv(&p);
|
|
398
|
11625 wbadword[badlen++] = 0;
|
|
324
|
11626 for (p = goodword, goodlen = 0; *p != NUL; )
|
|
474
|
11627 wgoodword[goodlen++] = mb_cptr2char_adv(&p);
|
|
398
|
11628 wgoodword[goodlen++] = 0;
|
|
324
|
11629 }
|
|
|
11630 else
|
|
|
11631 #endif
|
|
|
11632 {
|
|
|
11633 badlen = STRLEN(badword) + 1;
|
|
|
11634 goodlen = STRLEN(goodword) + 1;
|
|
|
11635 }
|
|
323
|
11636
|
|
|
11637 /* We use "cnt" as an array: CNT(badword_idx, goodword_idx). */
|
|
|
11638 #define CNT(a, b) cnt[(a) + (b) * (badlen + 1)]
|
|
|
11639 cnt = (int *)lalloc((long_u)(sizeof(int) * (badlen + 1) * (goodlen + 1)),
|
|
|
11640 TRUE);
|
|
324
|
11641 if (cnt == NULL)
|
|
|
11642 return 0; /* out of memory */
|
|
323
|
11643
|
|
|
11644 CNT(0, 0) = 0;
|
|
|
11645 for (j = 1; j <= goodlen; ++j)
|
|
|
11646 CNT(0, j) = CNT(0, j - 1) + SCORE_DEL;
|
|
|
11647
|
|
|
11648 for (i = 1; i <= badlen; ++i)
|
|
|
11649 {
|
|
|
11650 CNT(i, 0) = CNT(i - 1, 0) + SCORE_INS;
|
|
|
11651 for (j = 1; j <= goodlen; ++j)
|
|
|
11652 {
|
|
324
|
11653 #ifdef FEAT_MBYTE
|
|
|
11654 if (has_mbyte)
|
|
|
11655 {
|
|
|
11656 bc = wbadword[i - 1];
|
|
|
11657 gc = wgoodword[j - 1];
|
|
|
11658 }
|
|
|
11659 else
|
|
|
11660 #endif
|
|
|
11661 {
|
|
|
11662 bc = badword[i - 1];
|
|
|
11663 gc = goodword[j - 1];
|
|
|
11664 }
|
|
323
|
11665 if (bc == gc)
|
|
|
11666 CNT(i, j) = CNT(i - 1, j - 1);
|
|
|
11667 else
|
|
|
11668 {
|
|
|
11669 /* Use a better score when there is only a case difference. */
|
|
324
|
11670 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc))
|
|
323
|
11671 CNT(i, j) = SCORE_ICASE + CNT(i - 1, j - 1);
|
|
|
11672 else
|
|
|
11673 CNT(i, j) = SCORE_SUBST + CNT(i - 1, j - 1);
|
|
|
11674
|
|
324
|
11675 if (i > 1 && j > 1)
|
|
323
|
11676 {
|
|
324
|
11677 #ifdef FEAT_MBYTE
|
|
|
11678 if (has_mbyte)
|
|
|
11679 {
|
|
|
11680 pbc = wbadword[i - 2];
|
|
|
11681 pgc = wgoodword[j - 2];
|
|
|
11682 }
|
|
|
11683 else
|
|
|
11684 #endif
|
|
|
11685 {
|
|
|
11686 pbc = badword[i - 2];
|
|
|
11687 pgc = goodword[j - 2];
|
|
|
11688 }
|
|
|
11689 if (bc == pgc && pbc == gc)
|
|
|
11690 {
|
|
|
11691 t = SCORE_SWAP + CNT(i - 2, j - 2);
|
|
|
11692 if (t < CNT(i, j))
|
|
|
11693 CNT(i, j) = t;
|
|
|
11694 }
|
|
323
|
11695 }
|
|
|
11696 t = SCORE_DEL + CNT(i - 1, j);
|
|
|
11697 if (t < CNT(i, j))
|
|
|
11698 CNT(i, j) = t;
|
|
|
11699 t = SCORE_INS + CNT(i, j - 1);
|
|
|
11700 if (t < CNT(i, j))
|
|
|
11701 CNT(i, j) = t;
|
|
|
11702 }
|
|
|
11703 }
|
|
|
11704 }
|
|
344
|
11705
|
|
|
11706 i = CNT(badlen - 1, goodlen - 1);
|
|
|
11707 vim_free(cnt);
|
|
|
11708 return i;
|
|
323
|
11709 }
|
|
307
|
11710
|
|
351
|
11711 /*
|
|
|
11712 * ":spelldump"
|
|
|
11713 */
|
|
|
11714 /*ARGSUSED*/
|
|
|
11715 void
|
|
|
11716 ex_spelldump(eap)
|
|
|
11717 exarg_T *eap;
|
|
|
11718 {
|
|
|
11719 buf_T *buf = curbuf;
|
|
|
11720 langp_T *lp;
|
|
|
11721 slang_T *slang;
|
|
|
11722 idx_T arridx[MAXWLEN];
|
|
|
11723 int curi[MAXWLEN];
|
|
|
11724 char_u word[MAXWLEN];
|
|
|
11725 int c;
|
|
|
11726 char_u *byts;
|
|
|
11727 idx_T *idxs;
|
|
|
11728 linenr_T lnum = 0;
|
|
|
11729 int round;
|
|
|
11730 int depth;
|
|
|
11731 int n;
|
|
|
11732 int flags;
|
|
381
|
11733 char_u *region_names = NULL; /* region names being used */
|
|
|
11734 int do_region = TRUE; /* dump region names and numbers */
|
|
|
11735 char_u *p;
|
|
351
|
11736
|
|
|
11737 if (no_spell_checking())
|
|
|
11738 return;
|
|
|
11739
|
|
|
11740 /* Create a new empty buffer by splitting the window. */
|
|
|
11741 do_cmdline_cmd((char_u *)"new");
|
|
|
11742 if (!bufempty() || !buf_valid(buf))
|
|
|
11743 return;
|
|
|
11744
|
|
381
|
11745 /* Find out if we can support regions: All languages must support the same
|
|
|
11746 * regions or none at all. */
|
|
|
11747 for (lp = LANGP_ENTRY(buf->b_langp, 0); lp->lp_slang != NULL; ++lp)
|
|
|
11748 {
|
|
|
11749 p = lp->lp_slang->sl_regions;
|
|
|
11750 if (p[0] != 0)
|
|
|
11751 {
|
|
|
11752 if (region_names == NULL) /* first language with regions */
|
|
|
11753 region_names = p;
|
|
|
11754 else if (STRCMP(region_names, p) != 0)
|
|
|
11755 {
|
|
|
11756 do_region = FALSE; /* region names are different */
|
|
|
11757 break;
|
|
|
11758 }
|
|
|
11759 }
|
|
|
11760 }
|
|
|
11761
|
|
|
11762 if (do_region && region_names != NULL)
|
|
|
11763 {
|
|
|
11764 vim_snprintf((char *)IObuff, IOSIZE, "/regions=%s", region_names);
|
|
|
11765 ml_append(lnum++, IObuff, (colnr_T)0, FALSE);
|
|
|
11766 }
|
|
|
11767 else
|
|
|
11768 do_region = FALSE;
|
|
|
11769
|
|
|
11770 /*
|
|
|
11771 * Loop over all files loaded for the entries in 'spelllang'.
|
|
|
11772 */
|
|
351
|
11773 for (lp = LANGP_ENTRY(buf->b_langp, 0); lp->lp_slang != NULL; ++lp)
|
|
|
11774 {
|
|
|
11775 slang = lp->lp_slang;
|
|
|
11776
|
|
|
11777 vim_snprintf((char *)IObuff, IOSIZE, "# file: %s", slang->sl_fname);
|
|
|
11778 ml_append(lnum++, IObuff, (colnr_T)0, FALSE);
|
|
|
11779
|
|
|
11780 /* round 1: case-folded tree
|
|
|
11781 * round 2: keep-case tree */
|
|
|
11782 for (round = 1; round <= 2; ++round)
|
|
|
11783 {
|
|
|
11784 if (round == 1)
|
|
|
11785 {
|
|
|
11786 byts = slang->sl_fbyts;
|
|
|
11787 idxs = slang->sl_fidxs;
|
|
|
11788 }
|
|
|
11789 else
|
|
|
11790 {
|
|
|
11791 byts = slang->sl_kbyts;
|
|
|
11792 idxs = slang->sl_kidxs;
|
|
|
11793 }
|
|
|
11794 if (byts == NULL)
|
|
|
11795 continue; /* array is empty */
|
|
|
11796
|
|
|
11797 depth = 0;
|
|
|
11798 arridx[0] = 0;
|
|
|
11799 curi[0] = 1;
|
|
|
11800 while (depth >= 0 && !got_int)
|
|
|
11801 {
|
|
|
11802 if (curi[depth] > byts[arridx[depth]])
|
|
|
11803 {
|
|
|
11804 /* Done all bytes at this node, go up one level. */
|
|
|
11805 --depth;
|
|
|
11806 line_breakcheck();
|
|
|
11807 }
|
|
|
11808 else
|
|
|
11809 {
|
|
|
11810 /* Do one more byte at this node. */
|
|
|
11811 n = arridx[depth] + curi[depth];
|
|
|
11812 ++curi[depth];
|
|
|
11813 c = byts[n];
|
|
|
11814 if (c == 0)
|
|
|
11815 {
|
|
|
11816 /* End of word, deal with the word.
|
|
|
11817 * Don't use keep-case words in the fold-case tree,
|
|
|
11818 * they will appear in the keep-case tree.
|
|
|
11819 * Only use the word when the region matches. */
|
|
|
11820 flags = (int)idxs[n];
|
|
|
11821 if ((round == 2 || (flags & WF_KEEPCAP) == 0)
|
|
381
|
11822 && (do_region
|
|
|
11823 || (flags & WF_REGION) == 0
|
|
390
|
11824 || (((unsigned)flags >> 16)
|
|
351
|
11825 & lp->lp_region) != 0))
|
|
|
11826 {
|
|
|
11827 word[depth] = NUL;
|
|
381
|
11828 if (!do_region)
|
|
|
11829 flags &= ~WF_REGION;
|
|
355
|
11830
|
|
|
11831 /* Dump the basic word if there is no prefix or
|
|
|
11832 * when it's the first one. */
|
|
390
|
11833 c = (unsigned)flags >> 24;
|
|
355
|
11834 if (c == 0 || curi[depth] == 2)
|
|
|
11835 dump_word(word, round, flags, lnum++);
|
|
351
|
11836
|
|
|
11837 /* Apply the prefix, if there is one. */
|
|
355
|
11838 if (c != 0)
|
|
351
|
11839 lnum = apply_prefixes(slang, word, round,
|
|
|
11840 flags, lnum);
|
|
|
11841 }
|
|
|
11842 }
|
|
|
11843 else
|
|
|
11844 {
|
|
|
11845 /* Normal char, go one level deeper. */
|
|
|
11846 word[depth++] = c;
|
|
|
11847 arridx[depth] = idxs[n];
|
|
|
11848 curi[depth] = 1;
|
|
|
11849 }
|
|
|
11850 }
|
|
|
11851 }
|
|
|
11852 }
|
|
|
11853 }
|
|
|
11854
|
|
|
11855 /* Delete the empty line that we started with. */
|
|
|
11856 if (curbuf->b_ml.ml_line_count > 1)
|
|
|
11857 ml_delete(curbuf->b_ml.ml_line_count, FALSE);
|
|
|
11858
|
|
|
11859 redraw_later(NOT_VALID);
|
|
|
11860 }
|
|
|
11861
|
|
|
11862 /*
|
|
|
11863 * Dump one word: apply case modifications and append a line to the buffer.
|
|
|
11864 */
|
|
|
11865 static void
|
|
|
11866 dump_word(word, round, flags, lnum)
|
|
|
11867 char_u *word;
|
|
|
11868 int round;
|
|
|
11869 int flags;
|
|
|
11870 linenr_T lnum;
|
|
|
11871 {
|
|
|
11872 int keepcap = FALSE;
|
|
|
11873 char_u *p;
|
|
|
11874 char_u cword[MAXWLEN];
|
|
381
|
11875 char_u badword[MAXWLEN + 10];
|
|
|
11876 int i;
|
|
351
|
11877
|
|
|
11878 if (round == 1 && (flags & WF_CAPMASK) != 0)
|
|
|
11879 {
|
|
|
11880 /* Need to fix case according to "flags". */
|
|
|
11881 make_case_word(word, cword, flags);
|
|
|
11882 p = cword;
|
|
|
11883 }
|
|
|
11884 else
|
|
|
11885 {
|
|
|
11886 p = word;
|
|
389
|
11887 if (round == 2 && ((captype(word, NULL) & WF_KEEPCAP) == 0
|
|
|
11888 || (flags & WF_FIXCAP) != 0))
|
|
351
|
11889 keepcap = TRUE;
|
|
|
11890 }
|
|
|
11891
|
|
381
|
11892 /* Add flags and regions after a slash. */
|
|
|
11893 if ((flags & (WF_BANNED | WF_RARE | WF_REGION)) || keepcap)
|
|
|
11894 {
|
|
|
11895 STRCPY(badword, p);
|
|
|
11896 STRCAT(badword, "/");
|
|
351
|
11897 if (keepcap)
|
|
|
11898 STRCAT(badword, "=");
|
|
|
11899 if (flags & WF_BANNED)
|
|
|
11900 STRCAT(badword, "!");
|
|
|
11901 else if (flags & WF_RARE)
|
|
|
11902 STRCAT(badword, "?");
|
|
381
|
11903 if (flags & WF_REGION)
|
|
|
11904 for (i = 0; i < 7; ++i)
|
|
390
|
11905 if (flags & (0x10000 << i))
|
|
381
|
11906 sprintf((char *)badword + STRLEN(badword), "%d", i + 1);
|
|
351
|
11907 p = badword;
|
|
|
11908 }
|
|
|
11909
|
|
|
11910 ml_append(lnum, p, (colnr_T)0, FALSE);
|
|
|
11911 }
|
|
|
11912
|
|
|
11913 /*
|
|
372
|
11914 * For ":spelldump": Find matching prefixes for "word". Prepend each to
|
|
|
11915 * "word" and append a line to the buffer.
|
|
351
|
11916 * Return the updated line number.
|
|
|
11917 */
|
|
|
11918 static linenr_T
|
|
|
11919 apply_prefixes(slang, word, round, flags, startlnum)
|
|
|
11920 slang_T *slang;
|
|
|
11921 char_u *word; /* case-folded word */
|
|
|
11922 int round;
|
|
|
11923 int flags; /* flags with prefix ID */
|
|
|
11924 linenr_T startlnum;
|
|
|
11925 {
|
|
|
11926 idx_T arridx[MAXWLEN];
|
|
|
11927 int curi[MAXWLEN];
|
|
|
11928 char_u prefix[MAXWLEN];
|
|
455
|
11929 char_u word_up[MAXWLEN];
|
|
|
11930 int has_word_up = FALSE;
|
|
351
|
11931 int c;
|
|
|
11932 char_u *byts;
|
|
|
11933 idx_T *idxs;
|
|
|
11934 linenr_T lnum = startlnum;
|
|
|
11935 int depth;
|
|
|
11936 int n;
|
|
|
11937 int len;
|
|
|
11938 int i;
|
|
|
11939
|
|
455
|
11940 /* if the word starts with a lower-case letter make the word with an
|
|
|
11941 * upper-case letter in word_up[]. */
|
|
|
11942 c = PTR2CHAR(word);
|
|
|
11943 if (SPELL_TOUPPER(c) != c)
|
|
|
11944 {
|
|
|
11945 onecap_copy(word, word_up, TRUE);
|
|
|
11946 has_word_up = TRUE;
|
|
|
11947 }
|
|
|
11948
|
|
351
|
11949 byts = slang->sl_pbyts;
|
|
|
11950 idxs = slang->sl_pidxs;
|
|
|
11951 if (byts != NULL) /* array not is empty */
|
|
|
11952 {
|
|
|
11953 /*
|
|
|
11954 * Loop over all prefixes, building them byte-by-byte in prefix[].
|
|
390
|
11955 * When at the end of a prefix check that it supports "flags".
|
|
351
|
11956 */
|
|
|
11957 depth = 0;
|
|
|
11958 arridx[0] = 0;
|
|
|
11959 curi[0] = 1;
|
|
|
11960 while (depth >= 0 && !got_int)
|
|
|
11961 {
|
|
390
|
11962 n = arridx[depth];
|
|
|
11963 len = byts[n];
|
|
|
11964 if (curi[depth] > len)
|
|
351
|
11965 {
|
|
|
11966 /* Done all bytes at this node, go up one level. */
|
|
|
11967 --depth;
|
|
|
11968 line_breakcheck();
|
|
|
11969 }
|
|
|
11970 else
|
|
|
11971 {
|
|
|
11972 /* Do one more byte at this node. */
|
|
390
|
11973 n += curi[depth];
|
|
351
|
11974 ++curi[depth];
|
|
|
11975 c = byts[n];
|
|
|
11976 if (c == 0)
|
|
|
11977 {
|
|
|
11978 /* End of prefix, find out how many IDs there are. */
|
|
|
11979 for (i = 1; i < len; ++i)
|
|
|
11980 if (byts[n + i] != 0)
|
|
|
11981 break;
|
|
|
11982 curi[depth] += i - 1;
|
|
|
11983
|
|
455
|
11984 c = valid_word_prefix(i, n, flags, word, slang, FALSE);
|
|
|
11985 if (c != 0)
|
|
351
|
11986 {
|
|
376
|
11987 vim_strncpy(prefix + depth, word, MAXWLEN - depth - 1);
|
|
366
|
11988 dump_word(prefix, round,
|
|
455
|
11989 (c & WF_RAREPFX) ? (flags | WF_RARE)
|
|
366
|
11990 : flags, lnum++);
|
|
351
|
11991 }
|
|
455
|
11992
|
|
|
11993 /* Check for prefix that matches the word when the
|
|
|
11994 * first letter is upper-case, but only if the prefix has
|
|
|
11995 * a condition. */
|
|
|
11996 if (has_word_up)
|
|
|
11997 {
|
|
|
11998 c = valid_word_prefix(i, n, flags, word_up, slang,
|
|
|
11999 TRUE);
|
|
|
12000 if (c != 0)
|
|
|
12001 {
|
|
|
12002 vim_strncpy(prefix + depth, word_up,
|
|
|
12003 MAXWLEN - depth - 1);
|
|
|
12004 dump_word(prefix, round,
|
|
|
12005 (c & WF_RAREPFX) ? (flags | WF_RARE)
|
|
|
12006 : flags, lnum++);
|
|
|
12007 }
|
|
|
12008 }
|
|
351
|
12009 }
|
|
|
12010 else
|
|
|
12011 {
|
|
|
12012 /* Normal char, go one level deeper. */
|
|
|
12013 prefix[depth++] = c;
|
|
|
12014 arridx[depth] = idxs[n];
|
|
|
12015 curi[depth] = 1;
|
|
|
12016 }
|
|
|
12017 }
|
|
|
12018 }
|
|
|
12019 }
|
|
|
12020
|
|
|
12021 return lnum;
|
|
|
12022 }
|
|
|
12023
|
|
475
|
12024 #if defined(FEAT_INS_EXPAND) || defined(PROTO)
|
|
|
12025 static int spell_expand_need_cap;
|
|
|
12026
|
|
|
12027 /*
|
|
|
12028 * Find start of the word in front of the cursor. We don't check if it is
|
|
|
12029 * badly spelled, with completion we can only change the word in front of the
|
|
|
12030 * cursor.
|
|
|
12031 * Used for Insert mode completion CTRL-X ?.
|
|
|
12032 * Returns the column number of the word.
|
|
|
12033 */
|
|
|
12034 int
|
|
|
12035 spell_word_start(startcol)
|
|
|
12036 int startcol;
|
|
|
12037 {
|
|
|
12038 char_u *line;
|
|
|
12039 char_u *p;
|
|
|
12040 int col = 0;
|
|
|
12041
|
|
|
12042 if (no_spell_checking())
|
|
|
12043 return startcol;
|
|
|
12044
|
|
|
12045 /* Find a word character before "startcol". */
|
|
|
12046 line = ml_get_curline();
|
|
|
12047 for (p = line + startcol; p > line; )
|
|
|
12048 {
|
|
|
12049 mb_ptr_back(line, p);
|
|
|
12050 if (spell_iswordp_nmw(p))
|
|
|
12051 break;
|
|
|
12052 }
|
|
|
12053
|
|
|
12054 /* Go back to start of the word. */
|
|
|
12055 while (p > line)
|
|
|
12056 {
|
|
|
12057 col = p - line;
|
|
|
12058 mb_ptr_back(line, p);
|
|
|
12059 if (!spell_iswordp(p, curbuf))
|
|
|
12060 break;
|
|
|
12061 col = 0;
|
|
|
12062 }
|
|
|
12063
|
|
|
12064 /* Need to check for 'spellcapcheck' now, the word is removed before
|
|
|
12065 * expand_spelling() is called. Therefore the ugly global variable. */
|
|
|
12066 spell_expand_need_cap = check_need_cap(curwin->w_cursor.lnum, col);
|
|
|
12067
|
|
|
12068 return col;
|
|
|
12069 }
|
|
|
12070
|
|
|
12071 /*
|
|
|
12072 * Get list of spelling suggestions.
|
|
|
12073 * Used for Insert mode completion CTRL-X ?.
|
|
|
12074 * Returns the number of matches. The matches are in "matchp[]", array of
|
|
|
12075 * allocated strings.
|
|
|
12076 */
|
|
|
12077 /*ARGSUSED*/
|
|
|
12078 int
|
|
|
12079 expand_spelling(lnum, col, pat, matchp)
|
|
|
12080 linenr_T lnum;
|
|
|
12081 int col;
|
|
|
12082 char_u *pat;
|
|
|
12083 char_u ***matchp;
|
|
|
12084 {
|
|
|
12085 garray_T ga;
|
|
|
12086
|
|
|
12087 spell_suggest_list(&ga, pat, 100, spell_expand_need_cap);
|
|
|
12088 *matchp = ga.ga_data;
|
|
|
12089 return ga.ga_len;
|
|
|
12090 }
|
|
|
12091 #endif
|
|
|
12092
|
|
236
|
12093 #endif /* FEAT_SYN_HL */
|
